Assessing the Feasibility of Citizens’ Basic Income Pilots in Scotland: Final Report Prepared by the Citizens’ Basic Income Feasibility Study Steering Group June 2020 Contents

Project Team and Acknowledgements 4 Forewords by CBI Feasibility Study Councillor Group 5 Executive Summary 7 Report Outline 18 Section 1: Introduction to Citizens’ Basic Income (CBI) 19 1.1 What is a Citizens’ Basic Income (CBI)? 19 1.2 History of CBI 19 1.3 Arguments for and against CBI 20 Section 2: Background to the project 23 2.1 Local context and challenge 23 2.2 Establishment of CBI Steering Group 24 2.3 CBI Feasibility Fund 24 2.4 Feasibility Study Objectives and Outputs 25 2.5 Feasibility Study Governance 26 Section 3: Research and Evidence Overview 28 3.1 Overview of Published Evidence 28 3.2 Evidence from Contemporary International CBI Experiments 32 3.3 Worldwide Contemporary CBI Experiments 37 3.4 Evidence from Community Engagement and Surveys 40 3.5 Commissioned Research 43 Section 4: Rationale for Piloting and Approach to Assessing Feasibility 45 4.1 Insufficiency of evidence 45 4.2 Piloting versus Modelling 46 4.3 CBI Policy Schemes proposed elsewhere 47 4.4 Introduction to the Feasibility Framework 48 Section 5: Proposed Pilot Model 52 5.1 Overview and Rationale of CBI Model Option(s) 52 5.2 Summary of CBI Pilot Model Variables 52 5.3 CBI Pilot Design Principles 57 5.4 CBI Pilot Model Level(s) 57 5.5 Interaction with Social Security 59 5.6 Interaction with Tax System 61

2 5.7 Pilot Duration 62 5.8 Pilot Populations 62 5.9 Pilot Model Payment Options 64 Section 6: Evaluability Assessment 68 6.1 Theory of Change 69 6.2 Outcome measurement 72 6.3 Study design 74 6.4 Process Evaluation 85 6.5 Evaluation costs 87 6.6 Conclusions and recommendations 88 Section 7: Policy Pilot Governance and Research Ethics 95 7.1 Introduction 95 7.2 Implications of a control trial: protecting intervention and control groups from relative harms 96 7.3 Assessing and minimising the potential impacts on different groups 103 7.4 Protection of participants’ rights: Participation and informed consent 105 7.5 Governance recommendations 110 Section 8: Feasibility Assessment 112 8.1 Strategic Feasibility 112 8.2 Institutional Feasibility 116 8.3 Psychological feasibility 128 8.4 Behavioural Feasibility 130 8.5 Financial Feasibility 132 Section 9: Economic Modelling Findings 136 9.1 Introduction 136 9.2 Approach 136 9.3 Results 140 9.4 Economic Modelling: Discussion 147 Section 10: Feasibility Study Conclusions and Recommendations 149 10.1 Pilot model rationale and recommendation 149 10.2 Evaluation and research considerations 153 10.3 Discussion of overall feasibility 155 10.4 Economic Modelling Findings 157 10.5 Learning and discussion 159 References 163

3 Project Team and Acknowledgements

Project Steering Group:

Wendy Hearty, Project Manager, Improvement Service/Public Health Scotland Gerry McCartney, Steering Group Chair, Public Health Scotland Mhairi Paterson, Policy Officer, Improvement Service/North Ayrshire Council Chris Adams, City of Edinburgh Council Coryn Barclay, Fife Council Neil Craig, Public Health Scotland Nicola Elliot, City of Edinburgh Council Andrew McGuire, Improvement Service Julie McLachlan, North Ayrshire Council Fiona Myers, Public Health Scotland Paul Vaughan, Fife Council Andy White, Glasgow City Council

We are grateful to the Scottish Government representatives who have had input to the Steering Group for the duration of this project.

Our thanks go to the following individuals and group who have commented or had input at various stages of the development of this project:

Gerry Quinn, Glasgow City Council Mandy MacEwan, Fife Council Child Poverty Action Group research group Fraser of Allander/Manchester Metropolitan University/IPPR research collaboration group Marcia Gibson, MRC, University of Glasgow (evidence consultation and peer review) Peter Craig, MRC, University of Glasgow (formal peer review) Morag Treanor, Heriot Watt University (formal peer review) Jim Lewsey, University of Glasgow (formal peer review) Annie Miller, CBINS CBI Feasibility Study Stakeholder Group

4 Forewords by CBI Feasibility Study Councillor Group

Foreword by Councillor David Alexander and Councillor David Ross, Co-leaders, Fife Council

Fife welcomes the publishing of this report. Since the release of the Fairness Matters Report by the Fairer Fife Commission in 2015, the council and its partners have supported the piloting of basic income. The council renewed its commitment to piloting basic income after the local government elections and in the joint agreement between the administration parties. The cross-party support for undertaking this feasibility work shows that local councillors understand the suffering that vulnerability, inequality and precarious employment have on the lives of individuals, families and children. The impact of Covid-19 across our communities has also highlighted the fragility of people’s lives and the differing ability to respond to changing events. Our work through the pandemic has shown that over 100,000 people (nearly a third of the population) have vulnerabilities that would potentially require extra support, between ourselves and the voluntary sector over two thousand people a week are being assisted with basics such as food and energy. At this time it cannot be right that such a high proportion of the people that live in our communities cannot be sure that they can have the basics needed for living or the basis from which to develop and fulfil their ambitions. Universal Basic Income may offer a solution to some of these issues and we look forward to continuing to support the case for the feasibility work to help bring fairness across Fife.

Foreword by Councillor Ricky Bell, City Treasurer, Glasgow City Council

My Council has a long-standing interest in exploring the potential of a Citizens’ Basic Income to impact positively on the Citizens of Glasgow. We are keen to test its ability to address inequality and mitigate against poverty and deprivation. It is imperative that we consider new policy options, better designed than the current system and more equipped to improve living standards and quality of life.

The COVID-19 pandemic puts even more focus on the need for change and fuels the desire to find different and more effective responses to the many challenges we now face. There needs to be a secure financial platform that allows space and time for people to build instead of the continuation of an inadequate model of social protection that gives little flexibility. Going forward we must better value the contributions that people make and promote a fairer society and more sustainable existence.

This report is a significant contribution to this necessary debate and the specific consideration of the feasibility of a CBI pilot scheme in Scotland. It is important that all tiers of government embrace this opportunity and respond positively to the report’s recommendations.

Foreword by Councillor Joe Cullinane, Leader of North Ayrshire Council

Citizens’ Basic Income is a bold and radical policy idea which has seen a rapid increase in public interest not only Scotland and the UK, but worldwide, as a potential solution to reducing poverty and tackling economic insecurity. Levels of poverty and inequality are stubbornly high in some of our communities and we must consider innovative solutions if we want to create a fairer society centred on wellbeing.

5 This ground-breaking research explores the feasibility of implementing a pilot of Citizens’ Basic Income in Scotland. A key strength of the study lies in the robust and evidence-based approach.

Whilst a pilot of CBI is desirable, support from both the Scottish and UK Governments is needed to overcome the substantive and complex legislative and delivery barriers to piloting a CBI. I urge both Governments to actively consider this report and work with the project partners to take the steps needed to move towards a pilot of CBI.

Bold initiatives like CBI are needed now more than ever given the hardship and fragility exposed by Covid-19 and as we look to build back better and fairer.

Foreword by Councillor Cammy Day, Depute Leader, City of Edinburgh Council

Rising interest in new forms of social protection have been evident for a few years now, but the Coronavirus pandemic has brought with it a new tide of anxiety and insecurity for people across Scotland. Right now, we know that more people than ever are struggling to pay their bills, and that people who were already living on low incomes are suffering the most. We know too that a return to business as usual is not an option and that bold new ideas are needed to help people keep their heads above water both through current difficult times and a changed future landscape.

Citizens’ Basic Income is just such an idea, and we welcome the invaluable contribution this research has made to our understanding of its potential to improve wellbeing and reduce poverty. The challenges to implementing a pilot are substantial, but not insurmountable, and the potential benefits are significant. With this research we can be sure that Scotland is better prepared than any other country to move towards piloting the policy.

6 Executive Summary

Key Findings

• This report presents comprehensive and detailed research into the feasibility of a Citizens’ Basic Income (CBI) pilot in Scotland. It has been developed by a successful collaboration of local government and public health, supported by national government. • The CBI Steering Group recommends that a pilot study of CBI is undertaken in Scotland. A robust and well-designed pilot would provide an invaluable opportunity to test and evaluate the effects of CBI within the Scottish context. • The feasibility study provides important, evidence-based insights to determine the extent to which a nationwide CBI policy is possible within the Scottish context. However there are some areas where evidence is limited or does not exist, creating uncertainty around the specific impacts a CBI may have on people’s lives. A robust pilot and evaluation of CBI could resolve this uncertainty by generating improved evidence of the impact of a CBI on a person’s behaviour in a Scottish context; produce improved (but incomplete) evidence of the impact of a CBI on economic and community-level outcomes; allow testing of design and implementation features; and stimulate policy debate to strengthen political feasibility. • The CBI Steering Group has developed a proposed model for a CBI pilot and completed an evaluability assessment to consider the methods through which a pilot could be evaluated against designated outcomes of interest. The proposed pilot would aim to understand the impact of CBI on poverty, child poverty and unemployment, as well as health and financial wellbeing, and experience of the social security system. • Pilot model recommendations: − It is recommended that a pilot should be a randomised controlled study, with two study areas where the whole community receives a CBI (one receiving the high payment, the other receiving the low payment). − The preferred model for piloting a CBI pilot in Scotland is based on 5 key principles: universal (paid to all); unconditional (no requirement to search for work); individual (not paid to households, like Universal Credit); periodic (paid at regular intervals); and made as a cash payment. − The model proposes two levels of CBI payments - one ‘high level’ based on the Minimum Income Standard and one ‘low level’ more closely aligned with current benefit levels. For both levels of payment, suspension of a range of existing income-related benefits is proposed, while others, primarily related to disability, housing, childcare and limited capability for work would continue.

7 • Care should be taken that participants who are vulnerable and/or with low incomes should not experience detriment (financial or otherwise) compared to individuals not involved in the study. • Direct intervention costs of a pilot based on the proposed model (net of savings on benefits and pensions and excluding administration and evaluation costs) are approximately £186m over three years for a study including both low and high levels of CBI. • There is significant interest and value in exploring the potential benefits of CBI via piloting, however there are substantial challenges associated with institutional arrangements for a pilot. This research has determined it is not currently feasible for any one level of government alone to deliver a pilot model of a CBI as described by the Steering Group. There would be substantive and complex legislative, technical and delivery changes required to ensure that a CBI interacts with the existing social security system in a way that avoids detriment to those on benefits and lowest incomes. • The majority of social security benefits a CBI would need to interact with are reserved to the UK Government, particularly those in relation to housing, child-care support and other top-up payments. Within current welfare and tax governance arrangements, political will and support across all levels of government (local, Scottish and UK) including the Department for Work and Pensions (DWP) and HM Revenue and Customs (HMRC) would be required to overcome these challenges. Without such support, the feasibility of a pilot, that minimises detriment, would require legislative changes to be made. • We hope that the findings of the feasibility study contribute to the wider discussions on Scottish society, economy and wellbeing, as well as help identify actions which may support future ambitions in relation to basic income or social security reform.

This report presents the findings of research into the feasibility of a Citizens’ Basic Income (CBI) pilot in Scotland. We set out to design a CBI pilot to test the role of a CBI in reducing poverty in Scotland and explore the feasibility of local basic income pilots. We did this by gathering and synthesising CBI evidence across published research, engaging with relevant organisations, learning from contemporary pilots, community engagement and survey data, and commissioning new research to address evidence gaps. This provided the Steering Group with a robust evidence base on which to develop the proposed pilot model, alongside an evaluability assessment to assess the best way of measuring whether a CBI would deliver on the outcomes of interest.

Here we summarise the findings of the feasibility study, including economic modelling of the potential impacts on the wider economy in the longer term in a scenario where the policy was rolled out nationally, and draw out implications to inform future discussions about CBI, social security and the wider Scottish society and economy.

8 Background and Purpose of this Report

The feasibility project builds on earlier work undertaken by local authorities who were exploring the possible contribution of a CBI in reducing poverty and tackling inequalities. The work has been undertaken in collaboration by four local authorities (City of Edinburgh, Fife, Glasgow City and North Ayrshire), Public Health Scotland and the Improvement Service. Funding of £250,000 over 2 years was made available by the Scottish Government for this project. Representatives from each of the collaborating organisations were brought together to form the Citizens’ Basic Income Feasibility Study Steering Group.

This report has been prepared by the Steering Group to communicate the findings of the project. The group has been asked to consider the role of a CBI in reducing poverty by exploring the feasibility of conducting local pilots in Scotland. Specifically, this included details of the ethical, legislative, financial and practical implementation of conducting a pilot as well as its potential costs, benefits and savings.

We commissioned two pieces of research:

• Exploring the social security implications of a CBI pilot was carried out by the Child Poverty Action Group in Scotland. This work explored how a pilot study of CBI might impact on the pilot participants’ eligibility for other welfare benefits and associated ‘passported’ benefits. • Economic Modelling of the potential distributional and macroeconomic implications of a national roll out of CBI was led by the Fraser of Allander Institute at the University of Strathclyde in collaboration with the Institute for Public Policy Research (IPPR) Scotland and Manchester Metropolitan University.

There has been informative and constructive engagement with relevant civil servants from the Scottish and UK Governments, council officers and wider stakeholders to gather information and insights, particularly in relation to the institutional and legal aspects of a CBI pilot. This provided the Steering Group with the best available evidence base to develop a pilot model, consider how to evaluate and measure the impact of the pilot model, and to assess the overall feasibility of conducting a CBI pilot in Scotland. This final report builds on the evidence presented in our Interim Report published in November 2019.

What is a Citizens’ Basic Income (CBI)?

There are many different models of CBI, varying in the level of payment, eligibility and the degree to which it replaces and interacts with the existing social security and tax systems. The general concept is based on offering every individual, regardless of existing welfare benefits or earned income, an unconditional, regular payment. The Basic Income Earth Network (BIEN) have defined a CBI as having five essential criteria:

• Periodic: Paid at regular intervals (for example every month or fortnight), not as a one-off grant. • Cash Payment: Paid as an appropriate medium of exchange, allowing recipients to decide how to use it. It is not paid in kind or using vouchers. • Unconditional: Paid without a requirement to work or to demonstrate willingness to work. • Individual: Paid on an individual basis – and not, for instance, to households. • Universal: Paid to all, without means test.

9 Rationale for a Pilot

While CBI is not a new concept there has been recent rapid growth in public and political interest. Among its advocates a CBI is seen variously as a way of promoting social justice and equality, reducing poverty and income inequality, removing work disincentives, addressing job insecurity and increasing freedom to make choices. Critics variously view it as encouraging labour market withdrawal, promoting state dependency, diverting funds from those most in need, risking the removal of other social programmes, and potentially costly. Despite this interest, there remain many uncertainties about how a CBI would work within a Scottish or UK context, and how this would fit with, or require change in, our current tax and benefit systems.

The available evidence suggests that a CBI could impact on a wide range of employment, social and health outcomes, but the evidence base for CBI is largely drawn from other contexts and may not be directly applicable to Scotland today. Evidence on health impacts is mixed, with some positive effects on birth weight and mental health, but less evidence for other health outcomes. Similarly, impact on labour market participation is mixed, but overall employment impacts were smaller for men and greater for women with young children. There is an absence of evidence to assess the effects on long-term service use and wider economic impacts.

There are several contemporary CBI-type experiments underway or in planning around the world. However, as it currently stands, a CBI pilot in its purest form, where all characteristics outlined above have been met, has never been tested in any developed country.

In considering a proportional approach to supporting policymakers when judging the value of a CBI, we explored alternatives to a pilot, such as directly rolling out the policy or conducting economic modelling only. The feasibility study goes some way to determine the extent to which a nationwide CBI policy is possible within the Scottish context, however there are some areas where evidence is limited or does not exist, creating uncertainty around the impacts of CBI. Well-designed, local pilots of CBI might be able to address many of these gaps and generate new evidence for the Scottish context. Specifically, a robust pilot and evaluation of CBI would: generate improved evidence on the impacts of CBI in a Scottish context; allow testing of design and implementation features; and stimulate policy debate to better understand the political feasibility. The group concluded that whilst modelling the economic impact of a national rollout has a role to play in exploring broader and longer-term impact of a CBI, this approach alone would be insufficient to fill the gaps in knowledge about the impacts of a CBI.

The CBI Pilot Model

We propose a preferred model of CBI for piloting in Scotland. A three-year pilot would allow sufficient time for the realisation of short- and some medium-term outcomes. This should be preceded by a one-year preparation period to help mitigate delays which could constrain or compromise the pilot. The group recognises that the model of CBI and implications for other policies (especially tax policy) in a pilot scenario would be different if CBI were being implemented nationally across Scotland.

10 CBI Characteristics and Pilot Design Principles

CBI Characteristic Pilot Design Principle Cash Payment CBI in monetary form, paid by bank transfer or similar. Not paid in kind or as a voucher. Periodic (including payment type) Regular payment (weekly, fortnightly or monthly options) Given prospectively Individual Individual payments for adults Child payments to main parent/guardian, usually mother For adults without capacity, payment made to guardian Universal Total population (within saturation site) with no means- testing or restrictions by income, age or individual characteristics Unconditional No conditions or sanctions, CBI as a right.

Two levels of CBI payment are proposed. The high level is based on the 2018 Minimum Income Standard (MIS) produced by the Joseph Rowntree Foundation in order to have a model that is likely to be able to substantially reduce or eradicate poverty.

The second, lower, level of payment is more closely aligned with current benefit entitlements and provides an opportunity to test the effect of an unconditional income with relatively little change in the level of income for those previously receiving benefits.

Proposed Payment Levels

Age Range Low CBI Level (per week) High CBI Level (per week) 0 to 15 years £84.54 £120.48 (payment to main carer/ (payment to main carer/ parent) parent) 16 to 19 years £84.54 £213.59 20 to 24 years £57.90 £213.59 25 years to pension age £73.10 £213.59 Pension age £168.60 £195.90

We do not want to propose models of CBI for piloting that will lead to direct financial detriment for participants. This is very difficult to guarantee given that the current social security system is designed to identify financial need and vary payment accordingly, whilst a CBI is designed to be universal. In order to achieve this balance, our research has identified that certain benefits would need to be continued alongside a CBI. These are disability, work capability, housing and childcare benefits. It should be noted that for benefits retained alongside a CBI, some conditions associated with eligibility will inevitably remain. For example, entitlement for housing, childcare and disability support is currently based upon individuals meeting certain eligibility conditions. 11 Proposals on Interaction with Tax and Benefits For the duration of the study, pilot participants who would normally be in receipt of elements paid within Universal Credit, and premiums and additions within Pension Credit and legacy benefits should be able to claim these alongside a CBI: Specifically, those relating to disability, limited capability for work, housing, childcare and caring.

A preferred model would be to suspend participant access to the following benefit entitlementsi for the duration of the study:

• Income Support (Personal Allowance) • Income-based Jobseekers Allowance (Personal Allowance) • Income-related Employment and Support Allowance (Personal allowance) • Child Tax Credit (Family Element plus Child Element) • State Pension • Child Benefit • Carer’s Allowance (Basic Rate and Scottish Supplement) • Universal Credit: Standard allowance for Single person • Universal Credit: First child/subsequent child payments

For the purposes of means-tested benefit calculation for retained benefits, CBI payments should be disregarded as income.

We suggest that a CBI payment should be included in the calculation of income for tax purposes. This would mean that all taxable income (CBI and non-CBI) which is above the Personal Income Tax Allowance threshold (currently £12,500) would be taxed.

Evaluation and Governance

The case for a pilot study is to provide a means of reducing the uncertainties about the impact of the policy, including any unintended negative consequences and implementation issues and inform any further decisions about a national policy. We have developed a method of piloting that would maximise the resulting learning. A summary of the key recommendations is provided below: a) Any pilot should be a randomised controlled outcome study with two intervention arms (one receiving the high payment, the other receiving the low payment) where the whole community receives a CBI. This offers the best way of understanding the potential impact of a CBI on a range of social and economic outcomes. Within the pilot communities, everyone receiving the CBI would be invited to take part in the evaluation. b) Based on the theory of change, the outcomes should be changes in poverty, child poverty, unemployment, community level social and economic effects, health and well-being, and experience of the social security system. The final list of outcomes should be developed over time as the theory of change evolves according to emerging evidence and/or new areas of interest. i Suggested suspensions are set out in the context of the CBI characteristic of payments being made to the individual. It is not the intention that any couple rates would continue to be paid within these existing benefits. Instead, within this model, a couple would each receive the CBI payment.

12 c) The study should be delivered alongside a control group comprising a stratified random sample of the population drawn from the same sampling frame as the pilot communities. It is recommended the evaluation is complemented by a process evaluation exploring the mechanisms by which any change in outcomes came about, including any facilitators and barriers to the successful implementation of the CBI and any differential effects between different groups of recipients. d) People leaving and people entering the study area should be included in the pilot, with consideration given to various eligibility criteria to reduce the risk that differences in the availability of the CBI between study and neighbouring areas affect the outcomes by distorting decisions regarding uptake of employment, moving area etc. Children born in the study area during the study period should also be included in the pilot and receive the CBI.

The recommendations above specify what we believe to be the most robust evaluation possible. However, there are substantial challenges and important limitations of a pilot and outcome evaluation on the scale and using the design proposed in this report. Limitations of the proposed pilot and evaluation include: a) the limits on generalisability to other areas b) the limited scope to explore the role of contextual effects due to the limited range of contexts (e.g. geographical, social, economic) in which the CBI pilot would be implemented c) the risk that single sites might be contaminated by economic shocks peculiar to those areas d) some of the potential impacts of a CBI are unlikely to occur because of the time-limited, geographically-narrow focus of the proposed pilot and because the institutional barriers might prevent all the features of the CBI proposed being tested.

There are several recommendations associated with ethical considerations of progressing a pilot and evaluation as described here. Care should be taken to ensure that participants, especially those who are vulnerable and /or on low incomes, do not experience detriment (financial or otherwise). A transition strategy to support all pilot participants before, during and post pilot should be developed to manage the risks associated with transitioning on or off a pilot or following changes in participant circumstances. Prior to commencement of the study and evaluation, the appropriate measures to ensure ethical approval, data protection, and Equality Impact Assessment should be taken. Appropriate legal and procedural advice should be sought to ensure that the selection of intervention areas (and exclusion of other areas), and the mandating of people to participate in a pilot is procedurally fair and reasonable and within the legal competence of the policymaking body.

Feasibility Assessment and Learning

There are several, interdependent aspects of feasibility which the Steering Group have used to test the viability and achievability of a pilot. Considerations of pilot design, evaluation and governance notwithstanding, a CBI pilot must also be politically feasible in order to progress to implementation, and this is considered across four domains: psychological, behavioural, institutional, and strategic feasibility. Some aspects of financial feasibility are also considered.

Psychological Feasibility The analysis undertaken by the Steering Group found that although recent UK and local surveys

13 demonstrate net approval for the principle of CBI, support for CBI varies according to different population groups. For example, there is high support among young citizens, people who are unemployed, or on low incomes. There is net disapproval for some groups, including pension age residents, higher income groups and people who are self-employed. Public support also varies according to the method of funding a CBI, with greatest support for CBI models funded through general taxation targeted at high income groups. Although analysis provides a general indication of psychological feasibility for a CBI, it is important to note that public opinion on the model specified by the Steering Group has not been measured and support may therefore differ according to design features such as level of payment, communities included in the study, duration of a pilot and interactions with the tax and benefit systems.

Behavioural Feasibility The available relevant published evidence suggests that a CBI could impact on a wide range of social, employment and health outcomes. However, the current evidence base for CBI is mixed and there is a lack of evidence to assess the effects on long-term service use and wider economic impacts. The ways in which people respond to a CBI are largely in response to the model design and institutional arrangements. For example, in countries with CBI-like pilots, the political context and framing have had considerable impact on the pilot design and outcomes of interest. Evidence suggests that removing or reducing conditionality has the potential to make significant impacts on participant behaviour and health outcomes. A pilot would have the potential to assess more fully the changes in behaviours that would arise from a CBI in Scotland.

Institutional Feasibility Support would be required across a range of institutions to implement a pilot of CBI. Our proposed pilot model would require the full collaboration of the Department for Work and Pensions (DWP) and Her Majesty’s Revenue and Customs (HMRC). The Steering Group have engaged with the DWP and HMRC over the duration of the feasibility study to explore institutional opportunities and challenges. The required legislative and delivery competencies for a CBI pilot are reserved to the UK Government and, therefore, at present neither the Scottish Government or Local Authorities on their own could introduce a CBI. For example, there are substantial challenges to delivering a pilot which adequately tests all the principles of a CBI set out in our preferred model, while also ensuring pilot participants (particularly vulnerable and low-income groups) are not in detriment, financially or otherwise, as a result of participating in the study. Reducing the scale or scope of a CBI pilot, or amending the design of the pilot model, would not eliminate these challenges.

Addressing these challenges would require substantive primary legislation and regulation changes which currently are mainly the responsibility of the DWP and HMRC. Changes to legislation would be complex, time-consuming and possibly costly, and would therefore depend on significant political will and interest. Another challenge is the inability of the DWP’s existing IT infrastructure to administer a CBI and make the necessary changes to other benefit entitlements during and after any pilot. As a result, political engagement alongside further detailed discussion with both the DWP and HMRC would be essential to consider how key barriers to institutional feasibility could be overcome.

Financial Feasibility and Economic Modelling Findings A CBI pilot raises several questions regarding financial feasibility. These include the short-term costs of the pilot scheme if a pilot is able to go ahead in the future and the longer-term fiscal and macroeconomic implications if a CBI was rolled out nationally across Scotland. Trialling an

14 intervention such as CBI is likely to involve substantial spending as it is attempting to meet the basic living costs for an entire community for the duration of the pilot study.

We provide indicative estimates of the direct costs of CBI payments within the pilot model we propose. The direct intervention cost of a pilot based on the assumptions associated with our proposed model, net of savings on benefits and pensions, would be approximately £62m over 3 years for a sample size of 2,500 at the high level of CBI and £124.5m for a sample size of 14,600 at the low level of CBI. This would give a total of approximately £186m over three years for a study including both low and high levels of CBI. These costs do not include implementation and administration costs, which were not possible to estimate given the challenges around institutional feasibility.

However, it is important to note that these costs and the administration costs of the trial, could be partly offset by additional tax revenue as incomes are boosted and depending on the tax rules agreed in the context of the pilot. The precise amounts will vary, but it must be remembered that, if rolled out nationally, a CBI policy would replace some existing programmes with the intention of doing so in a fairer, more progressive and efficient way.

The economic modelling work commissioned and undertaken by the Fraser of Allander Institute and partners provides important insights into the potential impacts of a national rollout of CBI on the economy in the longer term. The modelling suggests that the CBI proposed could reduce poverty, child poverty and income inequality. It would also have the potential to reduce economic precarity and change the nature of labour/leisure/training/creating/caring choices that people currently face.

The modelling also, however, outlines the potential fiscal costs of a nationwide CBI scheme and the considerable uncertainty regarding the economic impact of both the CBI and the way in which it is funded. Overall, the modelling concludes that the CBI would have major impacts on society and the economy, driven by how citizens’ behaviour changes in response to the introduction of a CBI and any changes in taxation required to fund it, and by other economic impacts such as changes in productivity. The modelling suggests that in most scenarios, the potential benefits of the CBI proposed could come at a cost of reduced economic growth and reduced real incomes for the richest groups compared to what they would have been in the absence of a CBI.

Our report does not make a recommendation on how a nationwide CBI scheme should be funded as our focus is on the feasibility of a pilot scheme. It is acknowledged that there is an almost limitless number of permutations of funding sources. The taxation options alone are numerous including new taxes such as carbon taxes, tourist taxes or land value tax and also the closing of tax loopholes. The scope for administration cost savings from a national scheme that minimised bureaucracy and a potential realignment of public spending priorities are also highlighted by some advocates of CBI. It is likely that any government looking to introduce a CBI would consider the full range of options at their disposal.

Conclusions

We conclude that while there are divergent views on CBI across the Scottish and UK political spectrum, even amongst advocates there is little evidence of a consensus view of a preferred model and features of a basic income, or its relationship with the existing welfare state. This research has determined there is interest and value in exploring the potential benefits of CBI via piloting, however it is not currently feasible to progress to a pilot model of a CBI as described by the Steering Group, without strong support across all levels of government (local, Scottish and UK).

15 This is due to the substantive and complex legislative, technical and delivery challenges associated with institutional arrangements for a pilot. Political will and support across all levels of government and, in particular, the DWP and HMRC would be required to further understand and overcome these challenges and support a pilot and evaluation.

The feasibility study suggests several possible areas for next steps or further work, to be considered by interested parties: a) Feasibility study approach – the novel, systematic approach used here to assess aspects of political, ethical and financial feasibility alongside design and evaluation has been constructive and informative. This approach could be adapted and used to systematically explore other policies or interventions at design or in advance of implementation. b) Exploring knowledge gaps and findings from economic modelling – the 3-stage approach taken in the CBI economic modelling programme sets out new ways to approach future economic modelling. Further work to refine the assumptions underpinning this modelling could be undertaken, as well as adapting the approach for modelling other interventions and economic scenarios that might meet the goals of a CBI without some of the challenges and costs of the CBI proposed in this work. c) Providing detail on how a CBI may interact with the social security system – it would be essential to the feasibility of any CBI pilot that more technical work is carried out by Scottish and UK governments on benefit interactions which details realistic options to avoid detriment and allow important entitlements to continue alongside a CBI pilot. d) Further public engagement work – it is likely that any future exploration of CBI in Scotland would need to involve further engagement with the Scottish public, and specifically any local communities affected by a pilot CBI. e) Exploration of ethical issues – this process unearthed several ethical issues that should be explored in relation to both policy testing and implementation.

Given the evidence-based and collaborative approach to the study, we suggest partners involved continue to reflect on the learning and are active in the dissemination of the study findings both in the UK and further afield.

To build upon this learning, we ask that the UK Government, including the DWP and HMRC, actively consider these findings and work with partners to further explore the technical feasibility of implementing a desired pilot of CBI in Scotland.

Closing Remarks

The obstacles identified through the feasibility study pose significant challenges in progressing the exploration of CBI in Scotland, and indeed the rest of the United Kingdom. The practical difficulties in implementing a pilot alongside the current social security system would require substantial capacity and resources to address current legislative constraints. The levers to address these barriers sit within the UK government and neither the Scottish Government or Local Authorities on their own could implement a CBI. In the current landscape, any pilot would require considerable commitment from the UK Government to make the necessary legislative, technical and procedural changes required.

16 This is not to suggest that there would be no benefit in piloting CBI in Scotland. The potential trade-off between slower economic growth and income restraint for the richest groups alongside reductions in poverty, inequality and potential improved outcomes across health, wellbeing, education and social outcomes is a discussion that is already in progress. The possibilities of a CBI as a policy response to poverty, inequality, precarious employment and a focus on promoting wellbeing could provide useful insights to the debate. It is hoped that the findings of the feasibility study contribute to the wider discussion on Scottish society and the economy, as well as help identify actions which may support ambitions in relation to basic income or social security reform.

17 Report Outline

This report outlines the findings of the Scottish Basic Income Feasibility Study. The study aimed to design a Citizens’ Basic Income (CBI) pilot to test the role of a CBI in reducing poverty in Scotland and explore the feasibility of local basic income pilots.

It has been collaboratively produced for the Scottish Government by a partnership of four Scottish local authorities (City of Edinburgh Council, Fife Council, Glasgow City Council and North Ayrshire Council), along with Public Health Scotland and the Improvement Service.

The feasibility study explores whether and how a CBI pilot in Scotland would work, what it might look like, what outcomes it could achieve and how these could be measured.

This report takes the following format:

Section 1 introduces the concept of CBI, along with some of the arguments as to why it should be considered.

Section 2 covers the background to the feasibility study, including the objectives and governance of the study.

Section 3 provides an overview of the research and evidence gathered and synthesised over the course of the study, including specifically commissioned research.

Section 4 outlines the rationale for exploring a CBI pilot and covers schemes proposed elsewhere, gaps in evidence and introduces the feasibility framework for this project.

Section 5 outlines the proposed pilot model that has been assessed for feasibility.

Section 6 outlines the evaluability assessment undertaken alongside the development of the pilot proposals.

Section 7 describes the issues arising in relation to the ethics and governance of the pilot proposals.

Section 8 offers an assessment of the political, ethical and financial feasibility of the pilot proposals.

Section 9 discusses the findings from economic modelling of the pilot proposals and issues related to CBI policy implementation.

Section 10 brings all this evidence together to discuss learning from the feasibility study.

18 Section 1: Introduction to Citizens’ Basic Income (CBI)

1.1 What is a Citizens’ Basic Income (CBI)?

Citizens’ Basic Income (CBI) (also known as Universal Basic Income, Basic Income or a Citizens’ Income) is defined as a, “periodic cash payment unconditionally delivered to all on an individual basis, without means-test or work requirement”.1 The general concept is based on offering every individual, regardless of existing welfare benefits or earned income, an unconditional, regular payment.

CBI Characteristics

• Periodic: Paid at regular intervals (for example every month or fortnight), not as a one-off grant. • Cash Payment: Paid as an appropriate medium of exchange, allowing recipients to decide how to use it. It is not paid in kind or using vouchers. • Unconditional: Paid without a requirement to work or to demonstrate willingness to work. • Individual: Paid on an individual basis – and not, for instance, to households. • Universal: Paid to all, without means test.

Sources: BIEN 2020

While CBI is not a new concept there has been recent rapid growth in political and public interest and the concept has both its proponents and critics. Among its advocates a CBI is seen variously as a way of promoting social justice and equality, reducing poverty and income inequality, removing work disincentives, addressing job insecurity and increasing freedom to make choices. Critics variously view it as encouraging labour market withdrawal, promoting state dependency, diverting funds from those most in need, risking the removal of other social programmes, and being potentially costly. Despite this interest, there remain many uncertainties about how this would work within a Scottish or UK context, and how this would fit with, or require change in, our current tax and benefit systems.

1.2 History of CBI

The concept of providing citizens with a basic income can be dated back to various points in history, with some references as far back as Ancient Greece. In the 16th Century, Sir Thomas More envisaged a society with a basic income in his fictional novel Utopia. In the centuries that followed, the concept was picked up by various thinkers in North America and Europe, including Thomas Paine and Thomas Spence.6 In the mid-1980s, a group of international academics and advocates formed the Basic Income European (now Earth) Network (BIEN) which continues to gather interest and support from around the world.2

19 A combination of factors has broadened its appeal in recent times: rising inequality, widespread economic insecurity, precarious work, and the potential for labour to be displaced by technological change including automation and artificial intelligence.

Despite this interest, a full CBI has not yet been implemented in any country although there have been several pilots of interventions that meet at least some of the basic criteria for a CBI. There have been recent tests of different forms of CBI in numerous countries worldwide, including Finland, Canada and the Netherlands. Finland has published interim results from the first year of its pilot study.ii However, as yet there have been no comprehensive results published of tests of CBI in the UK or countries with similar welfare state provision.

The recent rapid growth in political interest has been fuelled in part by an organised international movement which has been researching and advocating for the policy for some decades. Civil society organisations and think tanks across the political spectrum, including the World Economic Forum,iii the Royal Society for the encouragement of Arts, Manufactures and Commerce (RSA),iv and the Adam Smith Institute,v have recently added their support to this movement.

There are many models of CBI. Differences between them include the CBI payment levels, how these might vary across different age groups, the source of funding, and the nature and size of reductions in other transfers that accompany it.

1.3 Arguments for and against CBI

Although there is growing political and public support for the concept, CBI is sometimes considered a controversial policy, with several arguments in favour and against (see Table 1). The validity and strength of arguments depend upon the specific design features of a CBI policy and the behavioural responses of recipients.

Proponents suggest a well-designed CBI policy has the potential to positively impact on many aspects of contemporary society by providing financial security and support for individuals whether they want to earn, learn, care or set up a business.3 Based on a rationale of reducing poverty and promoting greater equality, CBI is regarded as a way of removing barriers which prevent individuals from reaching their full potential.

ii Finland published preliminary results of its pilot (January 2017 – December 2018) in February and April 2019. Full results of the pilot will be available in 2020. https://www.kela.fi/web/en/basic-income- experiment iii https://www.weforum.org/agenda/2017/01/why-we-should-all-have-a-basic-income/ iv https://www.thersa.org/action-and-research/rsa-projects/economy-enterprise-manufacturing-folder/ basic-income v https://www.adamsmith.org/blog/nine-arguments-against-a-basic-income-system-debunked

20 Table 1: Commonly Used Arguments For and Against CBI

Arguments for CBI Arguments against CBI Promotes social justice and equality Encourages labour market withdrawal Reduces income variability, poverty & Promotes state dependency income inequality Increases liberty Unaffordable Requires raising tax levels to an untenable Removes work disincentives level Reduces complexity of social security system Diverts funds from those most in need Facilitates time for caring, education, Provides justification for removing other volunteering, arts, etc. social programmes Increases entrepreneurship Reduces job insecurity and in-work poverty Mitigates job loss due to automation

Recent statistics suggest 10.5% of workers in Scotland are in insecure employment, including those employed on zero-hour contracts, agency, casual or seasonal workers, and low-paid self-employed.4 The use of sanctions to suspend social security payments has also become more of an issue in the last 10 years.5 One of the primary supporting arguments for a regular CBI payment is improved income security and the associated benefits this provides to individuals in terms of predictability of income and lower income-related stress and anxiety.

Similarly, CBI could also provide financial security and support for individuals to engage in socially and personally productive activities such as community or voluntary work, caregiving, entrepreneurial activities, or creative pursuits.

Professor Guy Standing, CBI advocate and academic, proposes that:

“A basic income would allow more people, and not only the well-to-do, to pursue their passions. This would not only be personally satisfying but could yield big dividends for society, through the encouragement of entrepreneurship, through creative endeavour and through socially valuable pursuits at all levels”.6

The arts, culture and creative industries are an example of where a CBI could have substantial impact on individuals and broader society. In February 2019, Citizens’ Basic Income Network Scotland and the Scottish Artists Union hosted an event to discuss how a basic income could impact artists living and working in Scotland. Following the event Scottish artist Jenny Lupton shared her thoughts on what a basic income could mean for her:

“[a basic income] would be an amazing help for artists like me – it would take the stress out of worrying how to survive when the work is quiet… Stress and worry cause huge mental health issues, not least depression, and there is a general perception amongst non-artists that what we do isn’t a ‘real’ job anyway! If every single person received this money, there would be more successful artists out there and we could value ourselves better”.7 21 Within UK creative industries, individuals from working class origins are currently in the minority and face income and culture-related barriers to participation and employment.8 It is estimated there are currently only 18.2% of people from working class origins employed in music, performing and visual arts within the UK.8 A CBI providing a secure and regular income foundation could help overcome financial barriers, supporting aspiring artists, actors and others employed in arts and culture to pursue their passion and combat the under-representation of people from working class origins within the industry.

While basic income has gathered interest from around the world, there have been various criticisms of the concept and specific basic income schemes. The universal aspect of basic income is a common source of contention, with questions based on why rich people should be given a basic income when they do not require it. A counter argument to this suggests that it is more efficient to pay everyone and instate a more progressive form of taxation than we currently have such that people with higher incomes are not necessarily better off financially. There are also community and social benefits associated with the reduction of stigma from targeted and means-tested welfare benefits.9

It is often suggested that a basic income scheme would be unaffordable within the current economic context. Basic income represents a radically different approach to social security which would require a restructured and progressive tax system to finance it. As part of economic modelling of a CBI scheme, this report will provide further clarity on the financial feasibility of a CBI policy or pilot and the steps required to achieve this.

There are concerns that people in receipt of a basic income would stop working, resulting in labour market withdrawal and a dependency on the state. While there is little contemporary evidence from basic income type pilots or policies to suggest individuals would reduce employment,10, 11 a pilot in the Scottish context would help to provide more robust and relevant evidence.

22 Section 2: Background to the Project

2.1 Local context and challenge

Prior to this work and independently of each other, four local authorities in Scotland started to explore the idea of piloting a basic income. There is a common interest in reducing poverty and tackling inequality, and the role that a basic income might play in this.

2.1.1 Edinburgh Edinburgh is well recognised as an affluent and growing city, but also as a city with wide levels of inequality with some of the most deprived communities in Scotland. Within Edinburgh, the Council has set out a clear direction for action to reduce poverty, inequality, and their impacts on communities in Edinburgh. While the case for intervention is clear, there remain significant strategic challenges in improving the co-ordination and impact of measures to reduce poverty. As part of this wider poverty and inequality reduction agenda, in August 2017 a motion to City of Edinburgh Council noted the plans of three Scottish local authorities – Fife, North Ayrshire and Glasgow – to develop pilot schemes for a Citizens’ Basic Income in Scotland. The motion, approved by Council, agreed that the Council should join and work with these three authorities to develop a pilot scheme for a CBI in Scotland.

2.1.2 Fife In November 2015, the Fairer Fife Commission recommended that Fife Council should identify a town in Fife in which to test out a pilot of unconditional basic income. Fife’s Programme for Administration, May 2017, reinforced the council’s commitment to work with partners to establish a pilot Universal Basic Income Scheme in Fife.

2.1.3 Glasgow In January 2017 the Royal Society of Arts (RSA) gave a presentation on CBI to Glasgow’s Poverty Leadership Panel (PLP). Following this presentation, and the positive feedback it received, the council provided funding to carry out some community engagement with local groups and other interested stakeholders. Glasgow also held a sounding board meeting with other relevant organisations, such as the Big Lottery, trade unions, the Convention of Scottish Local Authorities (CoSLA), the Child Poverty Action Group (CPAG) and Glasgow’s Chamber of Commerce. In addition, two motions were approved by the full council. In November 2017, they resolved to reconvene a cross-party working group on trialling Universal Basic Income (UBI) for the city of Glasgow, noting the cross-party work done to date; and in June 2019, they resolved to continue to work to counter inequality by structural measures such as supporting progressive taxation and promoting universal basic income as an alternative social security system. The council’s strategic plan, which sets out the approach to improving the life chances of Glasgow citizens, includes a commitment to explore the feasibility of a basic income.

2.1.4 North Ayrshire As part of the budget-setting process in March 2017, North Ayrshire Council agreed funding of £200,000 for a basic income pilot which would look at the feasibility and potential benefits of implementing a basic income in North Ayrshire. In August 2017, North Ayrshire Council Cabinet agreed to permit officers to develop a feasibility study for a CBI pilot. The current 2019-2024

23 Council Plan includes a commitment to explore the feasibility of basic income pilots in Scotland. High levels of inequality and poverty exist in North Ayrshire. Unemployment levels in North Ayrshire are high, there are significant numbers of people on low income and almost a third of children live in poverty. This research is part of a wider package of support relating to the Community Planning Partnership’s Fair for All Strategy which aims to reduce inequalities in North Ayrshire. The council is committed to eradicating poverty and tackling inequalities within North Ayrshire. Exploring the feasibility of basic income pilots and their potential to reduce poverty and inequality is a way of challenging these issues.

2.2 Establishment of CBI Steering Group

The Steering Group was formed in November 2017, its first task being the submission of a joint application to the Scottish Government CBI Feasibility Fund. It was agreed that a joint bid between the four local authorities, supported by NHS Health Scotland and the Improvement Service, would be submitted to ensure research and expertise could be shared, to avoid duplication and to ensure the findings could be applicable to other local authorities in Scotland. Following this the Steering Group took on the role of managing and co-ordinating the work of the group and ensuring that the project was delivered as agreed.

The Steering Group included representation (from officers, civil servants and NHS staff) from each of the partner organisations:

• City of Edinburgh Council • Fife Council • Glasgow City Council • North Ayrshire Council • NHS Health Scotland • Improvement Service • Scottish Government

In addition to the £250,000 fund allocated by Scottish Government, each organisation provided in- kind resource to progress the feasibility study.

As agreed in the feasibility funding application, a Project Manager (based at the Improvement Service) and Policy Officer (funded by and based at North Ayrshire Council) were recruited in August 2018.

Several subgroups of the Steering Group with specific expertise were formed on an ad-hoc basis to take forward key elements of the feasibility study. This included a research sub-group convened to consider the published research and identify gaps that required commissioning of specific research projects, and an economic modelling sub-group to progress further research on economic modelling (further information in Section 3).

2.3 CBI Feasibility Fund

In September 2017, the Scottish Government announced in the Programme for Government that it would support local authority areas to explore the feasibility of a CBI Scheme. In March 2018, four local authority areas – Fife Council, City of Edinburgh Council, Glasgow City Council and North

24 Ayrshire Council – collaboratively prepared and submitted a joint bid to the Citizens’ CBI Feasibility Fund. The Scottish Government confirmed on 21 May 2018 they would provide £250,000 over two years to support the undertaking of a feasibility study for a CBI pilot in Scotland.

Ahead of this final feasibility report, the four local authorities – supported by NHS Health Scotland (now part of Public Health Scotland) and the Improvement Service – reported to the Scottish Government on the interim findings of the feasibility work in autumn 2019.

“Several Scottish local authorities are considering how they can pilot elements of a Citizens’ CBI, a radical form of social assistance. One of its attractions is that it may help those on the lowest incomes back into work or help them work more hours, while providing an unconditional ‘CBI’ as a safety net. We believe that bold and imaginative projects like this deserve support, but we also recognise that the concept is currently untested. Therefore, we will: • establish a fund to help these local authorities develop their proposals further and establish suitable testing • ask the Poverty and Inequality Commission to consider how it could help to draw together findings from local authorities to inform the government’s thinking.” Scottish Government, Programme for Government, September 2017

2.4 Feasibility Study Objectives and Outputs

Feedback from the interim report shaped the final pilot design options, evaluation considerations and final recommendations of the feasibility project. Supported by evidence on the ethical, legislative, financial and practical considerations associated with piloting a CBI, we have now considered whether piloting is feasible under current circumstances.

The Feasibility Study comprised five workstreams:

1. Project governance 2. Evaluability Assessment 3. Research 4. Engagement with relevant government departments 5. Communications and engagement

An evaluability assessment was undertaken, led by NHS Health Scotland who hosted four evaluation workshops which achieved:

• Agreement on the intended outcomes of a potential Scottish CBI pilot • Preferred options for CBI pilot/s to meet the intended outcomes • Identification of the intended and unintended consequences of the pilot option/s • Exploration of research questions to be tested through the pilot/s

25 Further to the evaluation workshops, two research projects were commissioned:

• Benefits-CBI Interaction research exploring how a pilot study of CBI might impact on the pilot participants’ eligibility for other welfare benefits and associated ‘passported’ benefits. This informed areas for negotiation with the DWP/HMRC and local authorities in relation to local arrangements for welfare payments/in kind benefits required to pilot the CBI. The report was published in June 2019.vi • Economic modelling of the distributional and macroeconomic implications of a CBI if it were implemented in Scotland. The findings of this work are published alongside this final report.

We engaged with the DWP and HMRC to explore the complexities surrounding the interaction of a potential pilot and the benefits system. We also engaged with communities in each of the four local authorities using a range of formats (further information on the feedback and findings from engagement is included within Section 3).

2.5 Feasibility Study Governance

2.5.1 Councillor Group As part of governance arrangements for the feasibility study, a Councillor group was formed comprising three cross-party representatives from each local authority (see Figure 1). The purpose of this group was to provide feedback on the progress of the Steering Group and give senior local authority input to the feasibility research programme.

As part of the governance arrangements, the initial Councillor Group meeting was held in March 2018, where terms of reference were agreed. A second Councillor meeting was held in October 2018, followed by other meetings in February 2019 and August 2019 to review the draft Interim Feasibility Report. The Councillor Group had further engagement in the process of agreeing the final Feasibility Report through April 2020.

Figure 1: Feasibility Study Governance

STEERING GROUP COUNCILLOR GROUP STAKEHOLDER GROUP Multi-disciplinary team from Provide feedback on the Feedback on plans at critical across councils, NHS Health progress of the steering group points in the development and Scotland, Improvement Service Give input to the feasibility study implementation of the study and Scottish Government and business case development Bring a range of views and Coordinate, develop, implement expertise and oversee the Scottish CBI Feasibility Study

March 2018; October 2018; February February 2018; November 2018; Meeting monthly from Nov 2017 2019; August 2019; May 2020 June 2019; November 2019

vi Report available at: https://basicincome.scot/wp-content/uploads/sites/75/2019/06/Exploring_the_ social_security-implications_of_a_basic_income_pilot_March2019.pdf 26 2.5.2 Stakeholder Group As part of the development of the feasibility study a Stakeholder group was established to provide a sounding board for the CBI feasibility project team and facilitate a flow of information between the team and interested groups and individuals. The open nature of this approach and the publishing of regular updates has established a clear communication route and maintained a useful dialogue with both advocates and sceptics of the policy.

The project has held four stakeholder workshops (in February 2018, November 2018, June 2019 and November 2019) which brought together a wide range of individuals and organisations to get broader feedback on the project plans. In addition to this, several meetings have taken place with the Poverty and Inequality Commission to update them on the study’s progress and seek feedback on specific areas of interest.

In autumn 2018 the group also participated in a series of workshops funded by the Scottish Universities Insight Institute (SUII), which brought together practitioners, policy makers and academics to explore a range of issues relating to basic income.vii

vii More information available at: https://www.scottishinsight.ac.uk/Programmes/Scotland2030/ BasicIncome.aspx 27 Section 3: Research and Evidence Overview

This chapter provides an overview of the evidence gathered to inform our proposed pilot model, feed into the evaluability assessment, and used to assess feasibility across different domains. The evidence spans previously published research, learning from contemporary pilots, community engagement and survey data, and commissioned research.

3.1 Overview of Published Evidence

Whilst a CBI has not been fully implemented anywhere, there are relevant studies from which we can learn about the potential impacts of CBI-type interventions on a range of outcomes. Studies which provide all individuals with an unconditional, substantial, regular sum of money may not be available, but there have been several studies that involve the unconditional provision of substantial cash payments to either individuals or households. To ensure maximum transferability of learning from findings, the focus is on evidence from upper middle- and high-income countries. This section provides an overview of the findings of these studies, along with consideration of the design, evaluation and impacts of these interventions. These studies encompass various types of interventions that can be broadly deemed as like a basic income: unconditional cash transfers (UCT), negative income tax (NIT), or those simply described as basic income (BI).

A systematic scoping review published by What Works Scotland10 synthesised evidence on studies which considered unconditional provision of cash payments to either individuals or households in middle- and high-income countries. A recent update of this scoping review was carried out, focusing on upper middle- and high-income countries only.11 These reviews provide a comprehensive overview of the evidence on basic-income type interventions where the context is most likely to be generalisable to Scotland. As such, a separate review of the available literature has not been systematically undertaken for this feasibility report. In addition, emerging evidence from contemporary basic income experiments from other middle- and high-income countries, such as Finland and Canada can be found in section 3.2.

The most recent review11 identified 27 studies of nine interventions conducted across a wide range of settings. These interventions were: the Alaska Permanent Dividend Fund (APDF); the Iranian Targeted Subsidy Plan (ITSP); various Native American Casino tribal dividend studies; the Ontario Basic Income Pilotviii (OBIP); and the North American Negative Income Tax (NIT) studies. The NIT studies included several sites – the New Jersey Graduated Work Incentive Experiment; the Rural Income Maintenance Experiment (RIME); the Gary Income Maintenance Experiment; the Seattle/ Denver Income Maintenance Experiment (SIME/DIME); the Manitoba Basic Income Maintenance Experiment (which included a saturation site in Dauphinix). The details of whether the interventions reviewed meet the characteristics of a CBI are summarised in Table 2.

viii This pilot began in 2018 but was terminated early upon change in the provincial government. More details on this pilot can be found in section 3.2.3 ix Although in principle the whole community of Dauphin was eligible to participate in the study, individuals only received a payment if their income dropped below the poverty level.

28 Table 2: Reviewed Study Intervention Characteristics

CBI Characteristic Intervention Unconditional Universal Permanent Basic Fixed* North American NIT studies (New Jersey, RIME, SIME/ Yes No No Yes No DIME, Manitoba) NIT saturation site (Dauphin) Yes Yes No Yes No Alaska Permanent Dividend Yes Yes Yes No Yes Fund Iranian Targeted Subsidy Plan Yes Yes Yes Yes Yes Tribal dividend studies Yes Yes Yes Varied Yes Ontario Basic Income Pilot Yes No No Yes No

*Amount does not change in response to other income

Whilst none of the interventions met all the characteristics of a CBI, some provided unconditional cash transfers to large numbers of people on a permanent basis, while others were time limited or smaller. The interventions were in different economic, social and political contexts, across different time periods, and some were targeted at specific populations. The existing welfare systems in the intervention countries varied widely and whilst the purpose of the interventions also varied, overall, they provided substantial additional income to individuals or households with no conditions attached.

The studies used a range of randomised control trials (RCTs), quasi-experimental, controlled before- and-after, and qualitative study designs. There were methodological issues with the studies, with several having small sample sizes, or poor reporting standards. However, some of the studies used more robust quasi-experimental methods and large samples. The studies provided evidence on labour market participation, education, health and wellbeing, social outcomes and spill over effects. The reported outcomes from the most recent review are summarised below.11

3.1.1 Labour market outcomes All of the studies reported impacts on labour market participation. In the NIT studies there were reductions in hours worked for male heads of household of between 1 and 9% reported, which primarily involved longer spells of job-seeking between employment. The review reports that absolute reductions in hours worked for second earners and female household heads were larger, up to a 33% reduction, but this was in contexts where no maternity leave was available. There was evidence to suggest that second earners and single parents spent more time in the home, and that the presence of preschool children in the home was a stronger predictor of hours worked for male heads of household than whether they received the NIT payment. Recent analysis of the Dauphin study (saturation site and part of the NIT studies) found a larger reduction in labour market participation than the other NIT sites and it is suggested that this was because people could respond to changing life circumstances and remain in work without losing benefits, and that this afforded greater autonomy and dignity.

29 In more contemporary studies (ADPF, Iranian, tribal dividend and OBIP studies), employment impacts were small overall. The evidence from Alaska is mixed, with one study reporting no long- term effects, and another a reduction in annual hours worked for both men (11%) and women (12%). Short-term labour market participation impacts were more mixed with one study showing a small increase for males. For women, there were reported small reductions in average weekly hours worked, particularly if they had young children, and in part driven by an increase in part-time working. Paid maternity leave is not generally available in Alaska. There was an increase in the probability of women working in the Iranian study.

Across the studies, the reported impact of providing unconditional payments on the labour supply evidence is inconsistent, but with mostly small impacts on men and larger impacts on women with young children. Where there are larger reductions in labour market activity the evidence suggests the time was directed into other productive activities.

3.1.2 Education Outcomes The NIT studies reported reductions in adolescent employment in favour of staying in education for longer, with substantial improvements in several educational outcomes for younger and more disadvantaged children reported in some studies, and with no impacts found in other studies. Qualitative evidence from Dauphin suggests that decisions to remain in school for teenagers were largely influenced by financial decisions, and in part by peer effects, which may have been driven by the saturation nature of this study site. A more recent tribal casino dividend study found an increase in educational participation for the most disadvantaged children.

3.1.3 Health and wellbeing outcomes Reported health effects were less consistent than those for labour market activity and educational participation, although this might be partly explained by differences in measurements used across the various studies or other study characteristics.

Health impacts were mixed across the NIT sites, with small improvements in psychological distress reported in some SIME/DIME sub-groups, but with no effects reported in New Jersey. In the tribal casino studies several analyses reported large positive effects on a range of child personality traits and mental health outcomes. There is evidence of a cumulative effect of exposure in some of the tribal casino studies, and effects were reported to be stronger in most high-risk groups. The study authors suggested that improved parental mental health due to decreased financial stress could have been the mechanism behind improved child mental health outcomes via improved supervision and better parent-child relationships.

Improved birthweight outcomes were reported for some of the studies including an improvement for high-risk groups in Gary and one study in Alaska (of up to 17.7g for each extra $1,000 in income). No effect on birthweight was found in the other NIT studies where this was measured. There were large reductions in infant obesity in one study of the Alaska Permanent Dividend Fund.

Health service use was measured in some studies. No impact was found in the New Jersey or SIME/DIME sites, but there was a large reduction in hospital admissions reported in Dauphin. No effects were found on a range of chronic conditions and health-related limitations in the New Jersey or SIME/DIME NIT studies, and there was no effect on mortality in SIME/DIME 40 years after the intervention.

Evidence from the tribal casino and Alaska Permanent Fund shows increased mortality rates

30 immediately following payments. In Alaska, between 2000 and 2006, a 13% increase in mortality is reported in the weeks following payments, with a reduction to normal levels four weeks later. In the Eastern Cherokee study, accidental death risk doubled following dividend receipt, with qualitative evidence suggesting that substance misuse and vehicle purchase are implicated, particularly following the first lump sum paid at age 18 years. It is worth noting here that payments in both Alaska and tribal casino dividends are paid as large lump sums once or twice a year and the latter can involve very large payments at age 18 years.

3.1.4 Social outcomes and spill over effects Some small positive effects were found across a range of social outcomes. No effects on marital dissolution were found in the NIT studies. Similarly, no effect on marital status was found in the tribal dividend studies but they did report positive effects on parental supervision, parent-child relationships and parental relationships. Similarly, mothers of small children in the Alaskan and Canadian studies reported spending more time at home with children. Where this data was collected, several studies reported small reductions in criminal behaviour, although in Alaska property crime decreased and substance abuse related crime increased in the month after payment receipt.

There is evidence to suggest that there could be important spill over and community level effectsx and it is possible that impacts on some outcomes could strengthen over time. For example, there is evidence of an increase in economic spending in Alaska, which in turn stimulated increased demand for labour. Improvements in mental health in the Dauphin saturation site seemed to benefit the whole community, although only 30% received payments. It is likely that reduced infant obesity in Alaska is linked to mothers spending more time in the home, and the projected resulting savings on health services are large. In addition, it is possible that the indirect effects of improvements in outcomes such as education and low birthweight can have widespread societal implications. The review also noted that there was evidence that improved mental health outcomes in the tribal dividend studies grew larger over time, suggesting that a permanent CBI could have greater long- term economic benefits than are evident in a shorter experiment.

3.1.5 Potential unintended consequences Some adverse impacts were reported, most notably, increases in mortality due to injuries (or accidents) related to receipt of transfers where these were larger, lump sum payments. Increased substance abuse was implicated in these increases, but it is noted that this pattern is observed after any large payments, including salaries and social security benefits.12 There were also some qualitative reports of increased substance abuse, particularly in the Native American casino studies. It is worth noting here that casino dividend schemes often result in very large lump sums for young adults which is likely to play a role in such substance abuse and in reported reductions in participation in further education or the labour market.

Cash transfers were linked to conflict in some communities, but this was related to issues around eligibility and resentment where there were targeted payments. This was particularly notable in the casino dividend schemes where receipt of dividends was dependent on being identified as a member of a native American tribe or community, even where other American citizens live locally and work in the same casinos. This clearly is very different contextually from Scotland so these specific unintended consequences may be less likely. However, we must be aware of the possibility x Spill over effects refer to impacts on individuals or communities who are not directly receiving the basic income payment

31 of conflict where communities adjacent to any pilot site may feel resentment toward those who are in receipt of a basic income.

There were also several implementation issues identified relating to perceptions of the intervention. The perception that a basic income would lead to recipients working less was persistent and led to negative publicity and political opposition in some cases. The Alaskan Permanent Fund is viewed more positively, reportedly because it is regarded as a common fund based on collective ownership of oil revenues.13

It is possible that a lack of interest in study findings from policymakers could have been due to poor understanding of the purpose, costs, and need for lengthy studies covering both intervention period and follow-up evaluation in CBI-type studies. In the case of the Dauphin study, this resulted in evaluation data being boxed up and not analysed for two decades. Conversely, public and political understanding of the purpose of social experimentation might have reduced controversy around some other experimental studies. Earlier NIT experiments suffered from a lack of co- operation between all relevant agencies and levels of government. Furthermore, targeted payments and issues establishing eligibility for payments were noted as possible causes for conflict and it was reported in the review that some recipients in such circumstances experienced hostility and media intrusion.

3.1.6 Conclusions The available evidence suggests that a CBI could impact on a wide range of employment, social and health outcomes, but the evidence base for CBI is largely drawn from other contexts and may not be directly applicable to Scotland today. Evidence on health impacts is mixed, with some positive effects on birth weight and mental health, but less evidence for other health outcomes. Similarly, impact on labour market participation is mixed, but overall employment impacts were small for men and greater for women with young children. There is an absence of evidence to assess the effects on long-term service use and wider economic impacts.

Impact on poverty levels is difficult to assess as this was not reported on as an outcome from the studies found. However, results from the tribal casino studies found that those children who were living in poverty at the start of the studies appeared to have benefited most from the dividend schemes. This was reflected in outcomes relating to child and adolescent mental health outcomes (with symptoms reaching the level of those children not previously living in poverty), and showing greater improvements in school attendance, length of time in education, and high school completion rates.

In addition to the evidence from published results of previous experiments or schemes, there are several contemporary CBI-type experiments underway or in planning around the world. The following section considers the evidence available from these activities and what can be applied to the Scottish context.

3.2 Evidence from Contemporary International CBI Experiments

There are several CBI-type experiments which have recently completed, are underway or are being planned around the world. Although specific to their local context, such experiments provide an opportunity to learn about the practicalities and challenges associated with implementing CBI pilots.

32 3.2.1 International Learning Report In August 2018, the Carnegie UK Trust funded a group of delegates from the Scottish Citizens’ Basic Income Steering group to participate in the 18th Basic Income Earth Network (BIEN) Congress in Tampere, Finland. Held over four days, the event brought together over 300 academics, policymakers and advocates to discuss and share opinions on a range of topics related to CBI. This provided a unique opportunity to engage with international colleagues working on developing, designing, implementing and evaluating pilots. Through our participation in the Congress, we were able to develop case studies on the areas where CBI-type pilots have recently taken place (Finland, the Netherlands, and Ontario, Canada) and have produced a report to share this learning.14

The report – Exploring the Practicalities of a Basic Income Pilot – provides background to the international study trip; explores why it might be useful to undertake a pilot of CBI; outlines the areas of feasibility the Steering Group are focusing on; presents case studies highlighting the key characteristics of areas where pilots are currently underway and highlights key insights from this activity that can inform the exploration of the feasibility of a CBI pilot in Scotland. The report outlined several considerations distilled from evidence in contemporary pilots:

• Context and framing have an impact on pilot design and outcomes of interest. For example, the pilots in Finland and the Netherlands were primarily interested in supporting people back into active participation in the labour market thus the pilots have focused on unemployed people and the outcomes of interest are employment.

• Connecting constituencies of support is necessary where support has grown from one specific organisation or arena. For example, in the Netherlands, interest first came from local authorities, so work was required to connect with national government. This is similar in Scotland where interest has emerged from Local Authorities and civic society – however, to deliver a feasible CBI pilot, collaboration is necessary with Scottish and UK governments.

• Political events and cycles have the potential to shape the design and future direction of pilots. In Finland, the pilot was made possible by a political window of opportunity, whereas in Ontario the pilot was terminated abruptly and early due to a change in political administration.

• Design is critical to the success of a pilot and early constraints can impact the pilot design. For example, in the Netherlands, federal legislation on conditionality constrained the municipalities’ ability to test alternative welfare designs so the experimental design has been compromised and some experimental conditions constrained.

• Conditionality has been highlighted by BIEN and the report case studies (and supported by work looking at Welfare Conditionality15) as having the potential to make significant impacts on participant behaviour and health outcomes. The tension between this fundamental design feature and current welfare policy and social norms is one of the biggest challenges in designing a feasible pilot.

• Measurement of impact on relevant outcomes is critical in designing pilots. In reference to the Ontario pilot, Professor Evelyn Forget noted that broad community outcomes (such as a reduction in access to health services, and migration in and out of the area) take more resources to evaluate and need to be planned early. We have placed importance on these outcomes and more information can be found in Section 6.

33 • Clear communications and public relations are critical in taking forward a pilot as there will be keen media attention if one were to take place. How the media report the impacts of pilots through ‘stories’ might overshadow the overall evaluation.

We were keen to focus on practical lessons for CBI pilots as much as possible rather than theoretical debate about the concept of a basic income. These lessons have helped inform the feasibility work in a variety of ways including the design of the pilot models and the assessment of strategic political feasibility (see Sections 5 and 8).

3.2.2 Preliminary Results from Finland’s CBI Experiment The Finnish basic income experiment commenced on 1st January 2017 and concluded on 31st December 2018. Preliminary results from Year 1 of the experiment were released on 8th February 2019.16

The interim results reported in this summary include preliminary register-based statistical analysis of the employment effects in 2017 and analysis of survey dataxi for the impact on the wellbeing of participants. Registry data for 2018 will not be available until later in 2020. These interim results are therefore only a partial reflection of the experiment and it is not possible to draw firm conclusions regarding the effects of basic income in this context.

Details of the experiment are outlined in Table 3.

Table 3: Summary of the Finland (2017-2018) Basic Income Experiment

Experiment Duration: 2 years (1st Jan 2017 to 31st Dec 2018) Experiment Governance: Approved by Government of Prime Minister Juha Sipilä Delivery Mechanism: Implemented and delivered by Kela, the Social Insurance Institution of Finland CBI Amount: $560 per month (equal to the basic unemployment allowance and labour market subsidy provided by Kela). Treatment Sample Size: 2000 individuals, aged 25-58 years who were in receipt of labour market subsidy or basic unemployment allowance in November 2016. Control Sample Size: 173,000 unemployed individuals. Selection Method: Selected via random sampling, with no geographical limitation. Participation was compulsory. Experiment Objectives: To study the effects of basic income on employment, income and wellbeing of participants. Evaluation Data: Data captured from both treatment and control groups: • Quantitative data from official registers • Telephone survey interviews (Oct – Dec 2018).

xi Due to time limitations on evaluation of the data, only some of the key survey results are reported in the preliminary results.

34 3.2.2.1 Preliminary Results – Effects on Employment and Benefits Quantitative data from official registers was collected from every person in the experiment (2,000 people in treatment group, 173,000 people in control group). Data collected included: days in employment, taxable income, participation in employment-promotion measures, and benefit receipt. The evaluation data is currently only available for the first year of the experiment (2017). Analysis of register data for the second year (2018) of the experiment will be available later in 2020.

Register data showed that during the first year of the experiment, basic income did not have an effect on the employment status of those who received. The intervention group was also no better or worse off in terms of obtaining employment than people in the control group. There is no measure of labour market withdrawal due to receiving basic income as only people in receipt of unemployment allowance (and therefore unemployed at the start of the experiment) were included in the study.

A critique of the study observes that some participants still faced a high degree of conditionality due to claiming other conditional benefits alongside receipt of CBI.17 These remaining conditional benefits meant that some people did not experience a less complicated benefits system and could still experience work disincentives as a result. In this sense, the study was not a true test of unconditionality, but instead tested a reduced number of conditional benefits.

With regards to the impact on the uptake of other benefits such as social assistance and sickness allowance, there was a noted difference between basic income recipients and the control group. Participants were still able to claim social assistance allowance while in receipt of basic income, however the basic income was considered as income in the calculation of social assistance. As a result, individuals receiving basic income received less in social assistance payments than the control group. However, there was no difference in the uptake of housing allowance between people in the treatment and control groups.

3.2.2.2 Preliminary Results – Effects on Wellbeing A telephone interview survey was conducted between October and December 2018. The survey targeted all participants of the treatment group and 5,000 people in the control group. The response rate was 31% for the treatment group and 20% for the control group, which made response bias highly likely. The survey questions asked about: trust, satisfaction with life, financial wellbeing, self-assessed physical and mental health, job-search activity, employment, and attitudes towards basic income.

The following summary provides an overview of the key indicators of different aspects of wellbeing as detailed in a report of the preliminary findings. The researchers evaluating the study advised that future reports would have more comprehensive analysis of the survey data including questions not reported in the preliminary results.

Compared to the control group, it was shown that the group receiving basic income had increased life satisfaction. Likewise, the confidence in one’s future, financial situation and ability to influence societal matters was greater for the basic income group than for the control group. Just over half of those receiving basic income (54.8%) and slightly less than half of the control group (46.2%) considered their state of physical and mental health to be good or very good. There was also less stress experienced by the treatment group than the control group. The researchers suggested that this may be due to the reduction in bureaucracy and a consistent flow of income afforded by basic income.

35 Respondents in the treatment and control groups were also asked about their trust in other people (generalised trust), or trust in institutions (the legal system or governments). The results showed that in comparison to the control group, the group receiving basic income demonstrated a slightly higher level of generalised trust. This was also true of trust in institutions, however the exact level of trust varied from one institution to another, with less trust in institutions which the public can directly influence (e.g. Parliament, political parties and the Government).

A recent blog by members of the Finnish pilot research and evaluation team hypothesised that increased levels of trust recorded by the treatment group could be because unconditional and universal policies are inherently based on an element of trust. Participants feel trusted, recognised and respected by the government and are consequently inclined to reciprocate feelings of trust. The researchers concluded that the, “relationship between trust and BI [basic income] is of the utmost relevance, and that enhancing trust should be a goal when reforming social security systems.”18

3.2.3 Results from Ontario, Canada In March 2016, the Ontario Liberal party committed to establishing a basic income pilot study to test whether it could sustainably reduce poverty. The experiment was described by the Ontario Provincial Government as a basic income pilot, however, it is more accurately described as a Negative Income Tax (NIT) model. Within this model, a minimum income was provided to participants living on a low income (under 34,000 Canadian Dollars (CAD) for a single person and 48,000 CAD for a couple).19 The level of income varied for each participant according to their additional earnings (income decreased by 0.50 CAD for every one CAD earned).

Despite this difference, the Ontario study is considered interesting for several reasons. As in Scotland, interest in the Canadian experiment is derived from a desire to tackle poverty and inequality. Furthermore, the study was initially piloted in three different locations, one of which was a semi-saturationxii site with the aim of measuring community-level effects. Owing to the premature cancellation of the study due to a change in the Ontario provincial government, the study also provides an opportunity to consider the effects of political influence and feasibility.

The cancellation of the project was announced in July 2018 after only 11 months of a planned 3-year pilot. Participants continued to receive payments until March 2019, however it is understood that no official evaluation has yet taken place. It is extremely unfortunate that the opportunity to collect data and evaluate the impact of the study was missed.

In lieu of an official evaluation, an advocacy group for the pilot study called Basic Income Canada Network (BICN) has undertaken a survey of some of the participants involved in the study. In their report, Signposts to Success,20 it is noted that 424 participants of the experiment provided responses to the survey (representing 9.4% of all 4,500 experiment participantsxiii). The survey asked participants a combination of closed and open questions based on their experience of the experiment. The questions included: their personal situation and demographics, level of education, pilot site enrolled in, employment status, and several questions asking how ‘basic income’ impacted on their future plans, finances, mobility, health and social inclusion. xii Any household which fell below a certain income level would be eligible for the basic income payment xiii Due to the difficulty in recruiting participants, the sample for the whole study was subject to selection bias. The qualitative sample will be even more biased as those who agreed to participate were systematically different from the main sample.

36 The BICN researchers were given access to official baseline data collected at the start of the experiment, however it was not possible to link baseline data with survey data. Despite this, the BICN researchers identified several themes derived from the survey data which offer reflections on how the income impacted on participants’ lives: a) The experiment suggested a positive impact on individual agency. Participant choices of how to use the income were based on unique personal circumstances such as paying down debts, improved dental care, returning to education or enrolling children in recreational programs. b) 88% of survey respondents suggested that the additional income had helped to reduce feelings of stress and anxiety. c) The survey responses suggest that social connections improved as a result of receiving a basic income. This is demonstrated through statements of increased time and financial ability to interact with friends and family, participation in volunteering opportunities and building social networks associated with enrolling in recreational activities. d) The experiment allowed recipients to make changes and improvements to their lives very quickly. e) 33% of the survey respondents reported using the income to invest in education or upgrading skills to provide better employment prospects. f) Themes around employment were associated with the multi-faceted, practical challenges associated with finding and maintaining good quality employment. Survey respondents noted a basic income helped pay for transport to work, such as bus tickets, fuel for cars, or essential car repairs. Reduced anxiety, improved health, better dental care and the ability to purchase new clothing was linked to increased confidence to pursue employment or education opportunities.

3.3 Worldwide Contemporary CBI Experiments

Table 4 provides a list of contemporary CBI-type experiments which have recently taken place or are currently underway around the world. These include experiments in higher and lower income countries and as such, have very different contexts and intervention designs.

Data on each of the experiments has been compiled to assess whether they meet the characteristics of a basic income. Some key design features such as the duration, pilot group size and annual payments are also provided.

Comparison of these contemporary experiments demonstrates that (according to available data) only the study in Uganda (and two intervention arms of the Kenyan pilotsxiv) meet all the criteria of a basic income. The Ugandan study is small in scale (provided to only one village with 56 adults and 88 children) and any associated learning may be limited for developed countries with established social security institutions. As it currently stands, a CBI pilot in its purest form, where all characteristics have been met, has never been tested in any high-income country.

Further details on each of these contemporary experiments can be found in Appendix 1. xiv The Kenyan study has three different intervention groups. One will provide a long-term monthly basic income to 44 villages for 12 years, another will give a short-term monthly basic income to 80 villages for 2 years, and the third will provide 71 villages with a short-term lump sum basic income.

37 Table 4: Contemporary Basic Income-Type Experiments Currently Underway, in Planning or Recently Completed Worldwidexv

Contemporary Equal & Test Group Annual Basic Cash Regular Unconditional Fixed Universal Duration Status pilot sites Individual Size payments Finland √ √ √ √ √ √ x 2,000 2 years €6,270 Completed. Evaluation underway Barcelona, √ √/x √ √/x x x x c2,000 3 years Varied Completed. Spain (local Evaluation currency) underway Gyeonggi √ √/x √ √ √ √ x c170,000 On-going 1 million Currently Province, South (local won ($883) underway Korea currency) Ontario, Canada √ √ √ √ x x x 4,000 3 years CAD Prematurely $16,989 ended July (s); CAD 2018. No $24,027 evaluation (c) (plus planned. additional CAD $6,000 for people with a disability) Oakland, √ √ √ √ − √ x 1,000 3 years US $12,000 In planning California, USA Kenya √ √ √/x √ √ √ x 21,000 2-12 US $273.75 Currently years underway Brazil √ √/x √ √ x − √/x c50,000 On-going 1560 In planning (local Mumbucas currency) (c. US $390) Uganda √ √ √ √ √ √ √ 1 village (56 2 years €200.40 Completed adults, 88 (per adult) children) xv Details correct as of November 2019 38 Contemporary Equal & Test Group Annual Basic Cash Regular Unconditional Fixed Universal Duration Status pilot sites Individual Size payments Netherlands √ √ √ √ x √ x 2,500 plus 2 years €11,352 Currently 500 (control) (individual), underway across six €16224 municipalities (family) Germany √/x √ √/x √ x √ x 250 3 years Varied Currently underway Stockton, √ √ √ √ √ √ x 125 18 US $6,000 Currently California, USA months underway

Key: ( - ) denotes missing / unknown data √/x partially meets criteria (s) Single household occupant (c) A couple

39 3.4 Evidence from Community Engagement & Surveys

The Steering Group considered a range of sources of evidence on public opinion as it currently relates to welfare policies and CBI. These include:

• Large scale international surveys assessing support for CBI across the UK and other developed countries. Within this, a key source of evidence has been the European Social Survey (ESS) which, in its 8th rotating round, incorporated questions on CBI schemes. Fieldwork for this research was carried out in 2016 and reported in 2017.21

• National surveys assessing views of, and support for, CBI within the UK and Scotland. Key sources for this element include an Ipsos MORI survey carried out for the University of Bath Institute for Policy Research in 2017,22 and a further survey carried out by Populus on behalf of the RSA in July 2018.23

• Local survey and focus group work assessing views of CBI with Scottish Local Authorities. In addition to these national and international survey results, during the past three years members of the Steering Group have taken forward focused analysis to assess the understanding and acceptance of CBI within the potential pilot local authorities. In particular, members of Fife People’s Panel (a representative panel of 1% of the adult population in Fife) were invited to complete either a paper or online survey between 16 June and 14 July 2017 incorporating questions on views of CBI.24 A similar approach was taken in North Ayrshire where, in 2018 the North Ayrshire People’s Panel (includes approximately 2,000 residents across the various communities of North Ayrshire) were posed similar questions on awareness and acceptance of CBI.25 During 2018/19 these local survey analyses were followed up with focus group work in both authorities.

3.4.1. Support for CBI Overall, analysis of these sources shows general net positive levels of support for CBI as a policy across all three levels of geography.

The results of the 2016/17 European Social Survey (ESS) found that 51% of UK respondents were ‘in favour’ of having a CBI scheme in their country, compared to 49% against. This result placed the UK at the mid-point of approval among European countries, with results ranging from 59% in favour in Belgium to 34% in Norway. Overall, only 12% of UK respondents were quoted as being ‘strongly against’ the policy.

Other UK level surveys conducted in recent years show even stronger levels of net approval. The 2017 Ipsos MORI survey found that 49% of respondents would support a CBI in the UK, compared to only 26% who would oppose the idea. The 2018 Populus survey found that 41% of adults would support the idea of a CBI in the UK, compared to only 17% who would oppose the idea in principle. Analysis of the results of this survey by UK region show similar levels of support recorded in Scotland, with 45% of adults in favour of the policy in principle and only 14% against (although this finding is based on a small sub-sample and should be treated with some caution).

When these national level surveys have been replicated for local authority areas in Scotland, the results remain positive. A survey of almost 900 people in Fife in 2017 found good level of public awareness of the concept of CBI, with 58% of respondents saying that they know something about it or understand it fully. Similar levels of awareness were also recorded in North Ayrshire where a 2018 survey found that 66% of respondents were aware of CBI as a concept.

40 When Fife residents were asked about how they would vote if a referendum was held in Scotland around the introduction of a CBI, respondents were split with 32% in favour, 29% against, and 39% unsure. A follow up question on support for a CBI trial in the area, however, brought a much clearer and positive response – 45% of respondents were in favour of a trial, with only 27% against. This result is in line with similar UK wide results found by the Populus survey which reported that 40% of respondents would support a trial in their local area, compared to only 15% who would oppose.

3.4.2 Perceptions of arguments given in favour or against a CBI When asked about their views on various arguments given in favour of a CBI scheme, UK-wide survey evidence by Ipsos MORI in 2017 showed support for arguments related to fairness, security, and transparency.

• Arguments that suggested CBI would be a way of rewarding important work that is unpaid, including care and other voluntary work were rated as convincing by 79% of all respondents to the 2017 Ipsos MORI survey. • In relation to security, 67% of respondents agreed that a CBI would provide a better safety net for those who felt insecure about their job. This was echoed by 63% who felt that CBI would be a good way of providing a guaranteed income if advances in technology and automation meant that people may lose their jobs and find it difficult to secure other employment. • Alongside these arguments, 62% of respondents agreed that the current welfare system is bureaucratic and that a CBI would reduce bureaucracy and make the system more transparent.

“It is universal - so would eradicate the demeaning process to access benefits. I have been shocked in the past how welfare staff treated me while applying and receiving benefits to which I was entitled.” Respondent to Fife Panel Survey

Local survey work carried out in Fife confirms many of these national level findings. The main argument that people find convincing in favour of a CBI is its perceived role in reducing anxiety about financing basic needs. This is followed by CBI encouraging independence and a sense of control, reducing bureaucracy and administrative costs, and creating more equality of opportunity.

In terms of arguments against a scheme, UK survey evidence points towards concerns over targeting, affordability, work incentives, and inflation, amongst other issues.

• 45% of all respondents to the 2018 Populus survey felt that money spent on giving everyone a CBI would be better targeted at the poor. This was a persuasive argument for two-thirds of all people who did not support a CBI, as well as for 40% of all those who were in favour of the scheme. • 42% of respondents felt that a CBI scheme would introduce disincentives to work. This was a particularly strong argument for opponents to the scheme, being cited by 72% of people who do not support a CBI. • 38% of respondents did not think that a basic scheme would be affordable, including 73% of all people who do not support a CBI, and • 23% felt that a CBI scheme would result in higher UK inflation. This represents a relatively low score overall but was an important argument for opponents to the scheme, being cited by almost half of those who do not support a CBI.

41 “People that are high earners don’t need it ... regardless of what they think ... Benefits should be for people in NEED.” Respondent to Fife Panel Survey

Again, local evidence corroborates many of these national findings. In survey work carried out in Fife the main arguments people found convincing against a CBI were that it would be difficult to finance, it would increase dependence on the state, and only the people who need it most should get something. Another reason given against CBI was that it would mean everyone paying higher taxes. Younger people appear to be more concerned about this. Less convincing reasons against CBI were that people should not be given something for nothing but should be linked to their efforts; and that others might come to the community to take advantage (of a CBI trial).

The Fife survey work also provides useful analysis on how local citizens felt that the introduction of a CBI might lead to changes in behaviour:

• Just under half (46%) of respondents said that a CBI would not affect how they would use their time. • 19% of respondents said they would gain additional skills, while 18% said they would look for a different job. Older people were more likely to say they would do more volunteering work. Younger people were more likely to choose to spend more time with family. • A small percentage (7%) said they would work for themselves. • Only 4% of respondents said they would work less than they do at present, with only 1% saying that the introduction of a CBI would lead to them stopping working. These tend to be people who are approaching retirement age.

3.4.3 Analysis by characteristic While survey data show general levels of net support for CBI, further analysis shows that these levels can vary considerably depending on the characteristics of respondents and the way the scheme is financed.

Data gathered as part of the 2016/17 ESS project allows for analysis of the drivers of support for CBI between different social groups. The analysis highlighted key areas of variance including employment status, income, age, and income source.

• Those who were unemployed at the time of the survey showed a 71% approval rate, compared to 53% approval amongst those who are in work. • Similarly, data showed a relationship between income level and support, with an approval rate of 64% among those who said they found it ‘very difficult’ to get by on their household income, against approval of 46% who felt ‘comfortable’ on their current income. • 61% of 15-34 year olds in the UK said they approved of the policy. Those aged over 65 years were the only age group in the UK who recorded a net disapproval rating, with only 43% being supportive of the policy. • In terms of sources of income, support is highest among those whose main source is unemployment benefits (65%), followed by those earning wages/salaries (about 57%). Support was lower among those receiving income from self-employment (49.5%) and farming (48%), and lowest for those receiving main income from investments/savings (about 37%).

42 At the same time as these variations, support for CBI also varied depending on how respondents thought the scheme would be financed.

• Data gathered by Ipsos MORI found that only 30% of respondents would support a CBI if it meant an increase in their taxes from current levels, compared to 40% who would oppose in those circumstances. • Overall, respondents would be most in favour of CBI if it were funded by raising taxes on the wealthiest in society. The Ipsos MORI survey found that 34% would prefer the policy to be funded by increasing taxes on wealth, compared to only 12% who supported it mainly being funded by an increase in general income tax. The Populus survey similarly found that 34% would support the policy being funded by raising progressive income tax so that the rich pay more than they got out of the scheme. There is also some evidence that support for CBI policy might be strongest when targeted at particular groups, although different surveys on this point do not show the same level of agreement as on other issues. The Ipsos MORI survey, for instance, found that 57% of respondents would support a scheme targeted at low income households, higher than the overall net support level of 49% for a genuinely universal scheme from that survey. The Populus survey, by contrast, showed no change in level of support between a universal or targeted scheme.

3.5 Commissioned Research

3.5.1 Benefits-CBI Interaction Research The purpose of this work was to explore how a pilot study of CBI might impact on the pilot participants’ eligibility for other welfare benefits and associated ‘passported’ benefits. The work considered the implications of different levels of options for the level of CBI, the potential for variation by age, and different responses by the Department for Work and Pensions (DWP), Her Majesty’s Revenue and Customs (HMRC), Scottish Government and Local Authorities to the CBI payments (i.e. whether or not these will be disregarded in assessing eligibility for other low income- related benefits and services).

The research considered the impact of the CBI pilot on all taxes paid and benefits received including, but not limited to, Universal Credit, housing benefits, disability benefits, passported benefits, and any local arrangements. The aim of this project was to inform the CBI pilot design considerations to help ensure that participants are in a position of ‘no financial detriment’. The findings have also informed discussions with the DWP, HMRC, Scottish Government, and local authorities in relation to locally arranged welfare payments/in kind benefits. The research project value was £10k and was commissioned in January 2019 to the Child Poverty Action Group in Scotland. The work was completed in March 2019 and the full report is available online.xvi

Further details of the findings of this project and how the findings have been used are found in Section 8 of this report.

xvi https://basicincome.scot/wp-content/uploads/sites/75/2019/06/Exploring_the_social_security- implications_of_a_basic_income_pilot_March2019.pdf

43 3.5.2 Economic modelling of the potential distributional and macroeconomic implications of a CBI The purpose of this commissioned work was to estimate through economic modelling the longer- term macroeconomic impacts of a CBI rolled out on a national basis. The research modelled low and high levels of CBI alongside changes in employment, tax revenues and savings to welfare benefit spending to estimate the full fiscal and economic effects of a Scotland-wide CBI. The study had three phases:

• Modelling the short-term effects of a CBI, in terms of labour supply response, and how these impact on the distribution of income across different income groups • Assessing the potential impact of these distributional effects on the macroeconomy through variables such as demand side changes in levels and patterns of spending, supply side changes in incentives and productivity, and what these changes mean for longer-term economic performance. • Finally, translating the macroeconomic effects into a second round of changes in the distribution of incomes across different income groups.

The research was led by the Fraser of Allander Institute at the University of Strathclyde in collaboration with the Institute for Public Policy Research (IPPR) Scotland and Manchester Metropolitan University. The approximate cost of this research project was £115k. The study reported in April 2020 and the full results are available online.26

A potential future stage of this research, which would require to be commissioned separately, would adapt and update the analysis based on the findings of the pilot studies, if these are progressed.xvii The pilot studies would provide insights on changes in some of the short-term outcomes likely to determine longer term economic outcomes, such as labour supply decisions. The findings from the pilot studies would not, however, be available for several years, even if they go ahead. Therefore, this commission is for the first stage only.

Further details of the findings of this project and how the findings have been used are found in Section 9 of this report.

xvii The current phase of work investigates the feasibility of conducting pilot studies in Scotland. A decision on whether to proceed to piloting will follow completion of that work.

44 Section 4: Rationale for Piloting and Approach to Assessing Feasibility

4.1 Insufficiency of evidence

Across the local authorities there is a common interest in reducing poverty and tackling inequality, and the role that a CBI might play in this. The local authorities recognise the potential of a CBI to support the delivery of existing strategies aimed at reducing poverty, unemployment and inequality. Our International Learning Report, which was informed by a study visit to the BIEN Congress in Finland, noted that despite substantial international and domestic interest in the concept, there remains doubt as to whether it would work within a Scottish or UK context, and especially within our current tax and benefit system and the norms of the social contract that system represents.14 Furthermore, there is currently insufficient consistent evidence on how people react and respond to CBI.9

Piloting CBI would have multiple benefits:

• Local pilots of CBI would enable small-scale, preliminary experiments to reduce the uncertainties about the impacts of the policy and thereby facilitate more evidence-informed policymaking • Unintentional and unpredictable consequences of CBI can be identified before large populations are exposed • A pilot encourages policy debate, helping to explore how and why CBI does or does not work27 • Pilots also have a role to play in raising awareness and generating public debate on the idea of a CBI.

The feasibility study goes some way to determine the extent to which a CBI pilot is possible within the Scottish context, however there are some areas where evidence is limited or does not exist. Well-designed, local pilots of CBI would be able to address these gaps and generate new evidence for the Scottish context.

The areas where learning from a pilot could contribute to the CBI debate or evidence base are listed below. a. Improved evidence of CBI on community-level impacts Community-level impacts are those which involve changed relationships or group-level organisations that can only be expected to occur if a whole community is exposed to CBI. For example, this could include the creation of new social enterprises to meet the needs of a community, informal childcare networks, or the creation of new clubs and organisations. There is currently limited relevant evidence into the community-level impacts associated with a CBI. Community-level impacts can only be measured via a saturation type model where everyone in a geographical area receives a CBI. Very few studies of this type have taken place or have been poorly evaluated for community-level impacts, resulting in an overall lack of evidence. A saturation type pilot would provide more exploration into the potential community impacts of a CBI, helping to address the lack of existing and relevant evidence in this area.

45 b. Improved evidence on impacts Hiilamo9 suggests there is currently insufficient consistent evidence on how people react and respond to CBI. This is particularly true of CBI pilots which universally provide a CBI, free of any employment related expectations or conditions. The academic and advocate of CBI, Guy Standing, notes that by focussing only on employment outcomes, some contemporary experiments are a “thinly veiled attempt to induce people to behave in ways the state thinks they should behave”.6 A pilot study which includes participants from a variety of backgrounds and measures a range of outcomes (either quantitatively or qualitatively) would provide enhanced evidence on how people respond to CBI.

There is also only sparse evidence on the likely impacts of CBI on the full range of primary and secondary outcomes of interest to us (see Section 6), including poverty, child poverty, employment, volunteering, entrepreneurial activity, health, security, etc. There is also the opportunity to identify unintended and unanticipated consequences of CBI before the whole population is exposed (which could be either positive or negative). c. Stimulate policy debate and strengthen political feasibility As a mechanism for evidence-based policy making, pilots contribute to the policy debate27 by helping to explore how and why CBI does nor does not work. They raise awareness of CBI both publicly and politically which when taken in conjunction with more certain evidence on impacts, could help strengthen political feasibility and legitimise rollout of a national policy if the impacts were found to be desirable. d. An opportunity to test design and implementation features A pilot provides an opportunity to test different policy design features. There are many models that can be used to pilot a CBI and the models piloted need to be chosen carefully and tested rigorously. Differences include the amount of the CBI delivered to participants; how payment levels are applied across different age groups; the source of funding; and the nature and size of reductions in other transfers that accompany it (for example, changes to tax and National Insurance systems; which benefits are withdrawn from participants). Testing a policy before rollout helps to identify potential implementation and design issues, providing an opportunity to amend design features before full rollout.

4.2 Piloting versus Modelling

It may be argued that the cost and time required to robustly pilot a CBI is not justified when modelling could yield useful evidence more quickly and cheaply. However, there remain several issues with modelling, which suggest that it would not be sufficient to robustly estimate the impacts of a CBI in Scotland. First, as with the evidence for CBI outcomes, the evidence on which to base to assumptions required for a model is sparse. Accurate, evidence-based estimations of behavioural responses to a CBI, particularly in relation to labour market participation, remain uncertain and this raises questions over the robustness of modelling outputs. Second, modelling would be insufficient to explore outcomes beyond primary outcomes related to labour market participation and poverty. There is interest in exploring the shorter- and medium-term outcome pathways by which longer- term outcomes might come about and these can only be explored more directly through piloting with a robust evaluation. Third, it is difficult for modelling to consider unanticipated and unintended consequences, or indeed implementation difficulties. These would be important considerations that would otherwise remain unknowns. It is therefore likely that a pilot would add substantially to the

46 evidence upon which policy decisions on whether to introduce a CBI for the whole population could be made.

4.3 CBI Policy Schemes proposed elsewhere

Since 2015, and from a variety of perspectives, multiple models for nationwide CBI schemes have been proposed that relate specifically to the UK, and more recently, the Scottish context. Policy schemes vary according to whether the level is set at a partial or full CBI, and whether the CBI is intended to fully or partially replace a number of current benefits. Housing, Disability and Carers benefits are commonly retained alongside a CBI. Typically, the payment levels proposed vary by age.

The Citizens Income Trust (2015) have described a model which is strongly grounded in the principles of a CBI: that it is paid on an individual basis, based on rights of citizenship, is unconditional, and not withdrawn as other income rises. It provides three different illustrative schemes for a CBI, two of which propose that a CBI should replace means-tested benefits, except for housing benefit and council tax support. The third proposes leaving means-tested benefits in place and taking a CBI into account. Under this scheme, the basic state pension and child benefit would continue to be paid. The levels of CBI would be indexed to average earnings, and savings from means-tested and other benefits could be used to fund a CBI.

These concepts are developed and articulated further in Annie Miller’s A Basic Income Handbook28 which informed early thinking around options for a pilot in Fife and Scotland, including as a minimum a CBI set at the level of means-tested benefits, and exploring a more generous basic income based on the EU Poverty Benchmark or a Minimum Income Standard for the UK.29

The Royal Society for Arts (RSA)30 proposed a basic income model of £71 a week for all qualifying citizens aged 25 to 65 years, a pension of £143 per week for those over 65 years, and variable payments for children based on age. While the adult level is on a par with means-tested benefits, the amount proposed for children and young adults is less than what low-income households would expect to receive under the current system.

The Buchanan Institute31 proposed modifications to the RSA model which would see a basic income being paid to all children (not just the first two), with young adults (18 to 24 years) receiving the full adult rate.

The Adam Smith Institute32 argued that existing welfare programmes are costly to administer and designed for a labour market that no longer exists. They propose replacing the majority of UK welfare benefits with a Negative Income Tax. This would be administered through the tax system, and act as a minimum income guarantee that would be tapered away as people’s earnings rise through work.

Similarly, Reform Scotland33 highlighted the need for reform of the current benefit system to ensure that work pays. Using proposals from the Scottish Greens (£100 per week per adult and £50 per week per child), they highlight which benefits would be replaced (such as out of work benefits and child benefit) and which should be retained (in relation to caring, disability and housing). Like the Citizens Income Trust, Reform Scotland proposes that a Citizens’ income could be funded in part by ending the tax-free personal allowance.

Compass34 explored the desirability and feasibility of introducing a basic income scheme in the UK.

47 Their first model proposes a full scheme to replace most means-tested benefits. The second is a modified scheme that would leave means-tested benefits in place, at least initially. They propose that it would be possible to implement a modified scheme based on a genuine unconditional income that would raise average incomes at the lower end of the scale and reduce poverty (particularly for children) and inequality. It could be implemented quickly and be a step towards a full basic income. They concluded that a full scheme that replaced all or most of the existing system would be difficult to implement in the current circumstances, it would be expensive, and there would be many losers among poorer households.

These models were extended in a further paper for Compass,35 which proposed a short-term scheme based on a partial basic income, and a long-term scheme based on a fuller basic income, funded through a Citizens’ wealth fund. The personal tax allowance would be converted into a cash payment to create a progressive form of universalism. The models would reduce poverty, inequality and reliance on means-tested benefits.

Malcolm Torry36 also explored different options for implementation including funding a basic income for all citizens through the existing tax and benefits system by maintaining means-testing but introducing new thresholds and considering an option that builds on child benefit and makes this universal for all ages over time.

The RSA37 proposal for a Universal Basic Income Opportunity Fund - £5,000 for all citizens for up to a two-year period - offered an alternative implementation option which would enable testing of the impact of the unconditional and social support aspects of a basic income.

IPPR Scotland38 included basic income as one of a number of options the Scottish Government could consider to address ambitious targets for reducing child poverty. The model of basic income used in the IPPR report was assessed to be a costly intervention that would increase relative poverty. IPPR proposed instead that a minimum income guarantee, a hybrid of a basic income and the current benefits system, would be less costly to implement.

Modelling by Landman Economics for RSA39 using the Scottish tax-transfer model illustrated that a partial basic income (£46 a week) would halve destitution and reduce relative household poverty by 8.5%, while a fuller basic income (£92 a week) would end destitution and reduce relative household poverty by a third.

More recently, Guy Standing40 proposed five different models for a basic income pilot, two of which propose saturation sites at different CBI levels, with and without means-testing being left in place. Other models recommended include payment of a common dividend as a supplement to existing benefits, removing conditionality from a sample of those in receipt of welfare benefits, and cash grants, for example, to the homeless.

4.4 Introduction to the Feasibility Framework

Debate around the concept of CBI often focuses around issues such as cost, the ethics of ‘giving money for nothing’, potential work disincentives, and the merits or weaknesses of previous pilot projects. However, the likelihood of an intervention becoming public policy is dependent on a wide range of influences and conditions, covering a range of complex and interdependent aspects of feasibility. For a policy to be successful, it must meet the aims and objectives set out in the policy intention – and achievable in the given context.41, 42 These two parameters underpin the analytical framework which the Steering Group have developed to assess the feasibility of a pilot scheme.

48 There are several, interlinking aspects of feasibility which the Steering Group have used to test the viability and achievability of a pilot. Ethical considerations and financial feasibility are two crucial aspects of this picture, but in relation to the feasibility specifically of a pilot, the basic income characteristics of the intervention to be tested and evaluated are equally important. Moreover, as argued by De Wispelaere and Noguera43 any proposed policy must also be politically feasible to allow either successful testing or roll out. Political feasibility is an overarching term to describe whether there is appetite for a particular policy, however to make sense of the wide range of political influences, it can be broken down into:Strategic , institutional, psychological and behavioural feasibility.43

To fully assess the feasibility of CBI pilots in Scotland, a framework encompassing all these interlinked concepts has been developed. This analytical framework has been used to test potential pilot CBI options across all aspects of feasibility in order that the preferred model reflects the most likely model which will adequately test key aspects of CBI, can be evaluated as such, is ethical, and will be politically feasible in order to provide robust evidence sufficient to consider the potential of CBI in Scotland. Figure 2 provides an overview of this analytical framework. Each aspect of feasibility is described in more detail below.

Figure 2: Feasibility Analysis Framework

FEASIBILITY MATRIX

Basic Income Characteristics

Achievability Viability (Prospective Constraints) (Retrospective Constraints)

Discrete Agency Strategic Institutional

Diuse Agency Psychological Behavioural

Financial

Evaluability

Ethical

Adapted from De Wispelaere & Noguera43

4.4.1 Political Feasibility De Wispelaere and Noguera43 described multiple aspects of political feasibility covering the interaction between different levels of agency or action, discrete and diffuse, with various potential constraints. Discrete agency refers to political power where there are readily identifiable individuals or groups with roles, responsibilities and intentions in relation to public policy, for example, political parties. Diffuse agency on the other hand is where political agency is exercised through no apparent coordination or collective intention, such as public opinion. Constraints may be

49 retrospective, such as the institutional infrastructure already in place, or the existing knowledge or beliefs of how individuals and groups are likely to respond to a policy. Constraints may also be prospective, such as the political will to build a coalition of support for a policy, or how it is received and understood by the general public. Political power and action interact with constraints to impact on the probability of a policy being implemented or the functioning and robustness of a policy once implemented. This gives rise to four interdependent domains of political feasibility:

1. Strategic feasibility is the strategic action to build a robust political coalition of support for enabling the legislation and subsequent adoption of CBI as public policy. This may involve politicians, political parties, social movements, interest groups, trade unions or other organised groups. Clearly different groups have varying levels of power and resource available to influence the policy making process, but it is not just the degree of their support that is of interest. Different groups may support different models of CBI, or have varying outcomes in mind for the policy, thus strategic feasibility is not straightforward and depends on the proposed CBI model and implementation levers required.

2. Institutional feasibility and context is important in terms of both political support and practical implementation issues. Political institutional support would be required from a range of organisations and this will impact on funding, administration and evaluation of a CBI pilot. Institutional feasibility affects desirability of a pilot and may affect long-term survival or political resilience of a CBI policy. Institutional feasibility is sensitive to strategic decisions over time as political strategy shapes the development of delivery institutions, thus these two aspects are closely linked.

3. Psychological feasibility is the idea of a CBI both readily understood and seen to be beneficial by the communities concerned? This encompasses public support of CBI and a key challenge here is the idea of reciprocity and contribution, the principle of ‘deservingness’ which has become the norm in the UK. Careful design and framing of a CBI may help influence public perception and thus support for piloting a CBI, but this must be done carefully to avoid undue impact on evaluation.

4. Behavioural feasibility concerns the behavioural changes due to the intervention that may affect the performance or survival of a policy in the longer term. Potential negative effects or unintended consequences of a CBI on individual behaviours must also be considered. This aspect is closely linked to psychological feasibility, particularly in relation to labour market behaviour, as expectations of behavioural impacts such as withdrawal from the labour market may undermine public opinion of the policy, thus in turn affecting strategic feasibility.

Each of these four aspects feed into overall political feasibility, which is complex, multifaceted and highly dynamic. Attempts to impact on one aspect of political feasibility is likely to interact with the others, as well as responding differently to different proposals relating to CBI models, piloting options, and evaluation plans.

4.4.2 Financial Feasibility The concept of financial feasibility relates both to questions regarding affordability as well as financial implications at an individual or household level. Cost of a CBI policy, or indeed pilot, will impact significantly on issues of strategic political viability, and financial impact on individuals have a bearing both on behavioural and psychological feasibility. Options to finance a CBI pilot or policy will be dependent on the design of the intervention, as well as the institutional context and

50 arrangements surrounding the pilot. The extent to which CBI implementation would impose financial losses on households or individuals is a crucial question for financial feasibility. The evidence to assess financial feasibility includes costings of specific proposed models, what funding and payment mechanisms can be negotiated with other parts of the system, including DWP and HMRC, as well as findings from modelling of both micro- and macro-economic impacts.

4.4.3 Evaluability A full evaluability assessment process has been undertaken to ensure that any pilot would be designed in such a way as to maximise the learning that would result. The main aim of the evaluability assessment is twofold: to assess to what extent can the outcomes of interest be tested within the context of a pilot, and; to what extent can these outcomes be robustly measured? The evaluability assessment involved clarifying with stakeholders the intended and unintended outcomes of CBI and assessing whether and how these could be measured. We have explored the nature of possible pilot models, their likely costs, the potential outcomes, and the hypotheses or research questions that the pilot will seek to address. The evaluability assessment process and findings are set out in more detail in Section 6.

4.4.4 Ethical Feasibility Ethical questions are numerous in relation to a CBI, and forms much of the debate around the concept. The main question for this feasibility study is whether a pilot of CBI can be delivered ethically in Scotland with the research ethics principle of ‘do no harm’ uppermost. A range of ethical considerations have been identified in relation to both the impact of the CBI policy in a Scottish context, and the evaluation of a pilot. Ethical considerations can impact on the acceptability of pilots as well as the potential costs of mitigating ethical issues, thus do not stand alone in assessing overall feasibility. Broadly speaking, the ethical issues relating to piloting a CBI include: consideration of mandatory participation; conditions required for ‘no detriment’; implications for participants post-intervention; issues of withholding a potentially beneficial intervention from controls; ensuring informed consent for the intervention and evaluation; incentivising the intervention and control groups to participate in the evaluation; and the implications of linking data from statutory organisations. A discussion of the full ethical considerations of evaluating and delivering a CBI pilot is set out in Sections 6 and 7.

51 Section 5: Proposed Pilot Model

5.1 Overview and Rationale of CBI Model Option(s)

A key aspect of the feasibility study is to make recommendations around how a pilot CBI could be designed and evaluated. Potential CBI models differ in terms of the size and characteristics of the eligible population(s),xviii the amount payable, the total cost and potential outcomes, all of which have implications for the evaluation design. The first part of this chapter (section 5.2) provides a summary and commentary on the variables for a pilot model design. The options preferred by the Steering Group are outlined and the rationale for these choices detailed in sections 5.3 to 5.9. A summary table of the preferred model is provided in Appendix 2.

Our model attempts to adhere to the overall principles of CBI whilst functioning alongside some existing social security arrangements to minimise the risks of financial detriment to vulnerable and low-income groups. We have also tried to find models that are more likely to be feasible for implementation in the current Scottish context given the shared responsibility for social security across the Scottish and UK parliaments and governments. However, it is difficult to find a model of CBI that would substantially simplify the social security system in this context whilst retaining sensitivity to the greater needs of some groups.

There is, from many, substantial criticism of the existing provision of social security. It was not our task to propose a new model of social protection, but instead explore the feasibility of a CBI pilot that would interact with elements of the existing system, whilst avoiding financial detriment to participants. However, our model does avoid some elements of the existing system that have been the subject of criticism (e.g. the two-child policy in universal credit and child tax credits, partial uprating of payments, five week waiting times, and sanctions associated with conditionality).

5.2 Summary of CBI Pilot Model Variables

There are a range of variables associated with options for a CBI pilot. These include: the amount payable, the characteristics of eligible populations, the number of people getting the CBI, the number of intervention sites, and the length of a pilot. These variables will have implications on the cost of a pilot and the evaluation design. The effect of the preferred model options on the evaluation design is outlined in Section 6.

Table 5 summarises the potential variations which could be introduced to a CBI pilot study. A commentary considers the pros and cons and where relevant, notes the rationale and recommendation of the Steering Group.

There are a number of ways in which a pilot could be designed, and some of these would not meet all characteristics of a CBI. The intention of the Steering Group has always been to adopt as pure a model of a basic income as possible, in order to test out these principles for the Scottish context.

We propose ‘design principles’ which we considered fundamental to robustly testing and evaluating a pilot of CBI. These pilot design principles are closely linked to the key characteristics of a CBI and xviii For example a saturation site where the all the eligible population within a geographic area receives the CBI, or a targeted study where only certain groups receive a CBI (e.g. people who are unemployed, care leavers, etc.).

52 are outlined in the following section. The rationale for the preferred CBI pilot model is based on adherence to the design principles and therefore evaluate the effectiveness of a true CBI pilot.

53 Table 5: Possible Design Variations Which Could be Introduced To a CBI Pilot Study

Pilot Design Variables Comments on Variables Preferred Option and Rationale Saturation versus Saturation: It is possible to have either a saturation or targeted pilot targeted pilot A saturation pilot is where all residents of a geographical study, however only a saturation study would test the area are included in a pilot study. A saturation pilot effects of a universal pilot of CBI. where all individuals receive a CBI would be expected A targeted pilot would undermine the universality principle to generate community-level effects associated with and would therefore not meet the design principles for a universality, which could be measured within this type of CBI. study. A targeted pilot would provide learning only for a specific Targeted: group of people and consequently have limited learning A targeted pilot is where participants are identified for the wider population. It would prevent community-level according to a shared characteristic and may be impacts associated with universality from being detected. geographically dispersed. Within a targeted study, it Steering Group recommendation: saturation model would be possible to identify a vulnerable group (e.g. care-experienced individuals, carers, lone parents) in order to consider the specific impact of a CBI on that particular group. Identification of both the intervention and comparison groups is very challenging for this model, but not impossible. Community-level impacts would not be detectable. Identification of differential impacts for vulnerable groups within a larger pilot would only be possible for those groups who are relatively large (e.g. women, possibly those with disabilities). High level CBI payment It is possible to model one or more CBI payment level and The Steering Group proposes there is learning value in versus low level CBI compare the effects of these. testing both payment levels and comparing the effects of payment A high level of CBI has potential to have a greater these. impact on poverty compared to a lower rate due to the Steering Group recommendation: test two payment increase in income. A low CBI at a similar level to current levels. In order to achieve this and a saturation approach entitlements provides an opportunity to test the effect of this would require two intervention sites – one for low CBI an unconditional income with no change in income level. and one for high CBI.

54 Pilot Design Variables Comments on Variables Preferred Option and Rationale High deprivation It would be possible to target communities with different The additional learning from pilots in different areas communities versus levels of deprivation and compare the impacts. However, would be limited but the costs would double. There is communities similar to the effects on sample size if we wanted to have enough also the opportunity to look at differential impacts by Scotland overall statistical power to be able to draw robust conclusions socioeconomic position within more average areas about differences in impacts would make the costs and because these will still have social gradients within them. scale of the pilots prohibitive. Steering Group recommendation: pilot in communities similar to Scotland overall. Urban communities It would be possible to target urban and/or rural The additional learning from pilots in different areas would versus rural communities communities. It is likely that the scale and type of be limited but the costs would double. community effects will differ between towns and rural Steering Group recommendation: do not pilot separately areas and it may be of interest to target both and compare in urban and rural areas. the effects. However, the effects on sample size if we wanted to have enough statistical power to be able to draw robust conclusions about differences in impacts would make the costs and scale of the pilots prohibitive. Large versus smaller Using Intermediate Zones as a basis would suggest A pilot site needs to be of a reasonable size in order community size a minimum community size of 2,500 individuals for to detect community-level effects. Research by the inclusion in the pilot study. Defining larger community Steering Group suggests Intermediate Zones (a statistical sizes depends on the CBI level in question and the geography comprising between 2,500 – 6,000 residents) corresponding statistical power required to detect would provide a suitable basis to identify intervention changes in the primary outcomes (poverty, child poverty sites. and unemployment). Larger community sizes increase Steering Group recommendation: the minimum size of the costs of a pilot, but there is little evidence available intervention site should be 2,500 people. to guide what population size community level impacts operate at.

55 Pilot Design Variables Comments on Variables Preferred Option and Rationale Duration of pilot It would be possible to vary the length of the proposed A three year pilot would allow sufficient time for the pilot. A pilot of less than two years may prevent the realisation of short and some medium-term outcomes. A realisation of some outcomes of interest which take longer one year preparation period would help mitigate delays to develop. A pilot greater than three years could be at which could constrain or compromise the pilot. risk of weakened political commitment if there was a Steering Group recommendation: a three-year pilot with change in government or a shift in policy approach. additional one-year preparatory period.

56 5.3 CBI Pilot Design Principles

The starting point for developing the CBI pilot model proposed are the fundamental characteristics of a basic income. These characteristics define the key design features of a CBI. The characteristics of a CBI are set out in Table 6 below.

The proposed CBI pilot models are designed in such a way as to ensure that each of the fundamental characteristics of a CBI can be tested as far as possible within a pilot context. These characteristics have already been discussed in Section 1 so will not be described in further detail here.

Table 6: CBI Characteristics and Pilot Design Principles

CBI Characteristic Pilot Design Principle Cash Payment CBI in monetary form, paid by bank transfer or similar. Not paid in kind or as a voucher. Periodic (including payment type) Regular payment (weekly, fortnightly or monthly options). Given prospectively. Individual Individual payments for adults. Child payments to main parent/guardian, usually mother. For adults without capacity, payment made to guardian. Universal Total population (within saturation site) with no means- testing or restrictions by income, age or individual characteristics. Unconditional No conditions or sanctions, CBI as a right.

Our international learning report14 highlights that comparative basic income schemes in Finland, the Netherlands and Ontario (countries with a similarly developed economy and welfare system to Scotland and the UK) meet the cash payment and regularity principles of a basic income. However, compromises were made in the design of some schemes reflecting the different political and institutional contexts operating in each country. For example, participants in Finland were restricted to those who were in receipt of unemployment benefits, so the scheme could not be deemed to be universal. In both Ontario and the Netherlands, payments were means-tested based on household income, so did not meet the basic income principles of individual and unconditional payment. An important element of the feasibility work is to explore whether we can remain true to these principles in designing and piloting a model of basic income for Scotland.

5.4 CBI Pilot Model Level(s)

Having considered many different levels of CBI, we propose two levels of CBI payment. For both of these payment levels the amount of CBI given varies according to age (albeit age bands differ slightly for the low and high payment levels). Advocates suggest that in a test of a pure CBI, age is the only permitted instance where the CBI rate varies according to personal attribute.28

One of the key intended outcomes of CBI is to reduce poverty. The High-Level CBI Payment is based on the 2018 Minimum Income Standard (MIS) defined by the Joseph Rowntree Foundation

57 (JRF) (Table 7).xix The rationale for testing a high level of CBI set at the MIS is based on the potential for a higher CBI to have a greater impact on poverty than a lower rate. The MIS defines a poverty threshold based on the income required to pay for, “items that members of the public think UK households need to be able to afford in order to meet material needs such as food, clothing and shelter, as well as to have the opportunities and choices required to participate in society.”xx

Table 7: Proposed High Level CBI Payment

Age Range Payment Rate (per week) Basis for CBI Rate 0 to 15 years £120.48 (payment to main carer / Based on rate for a primary parent) school-aged child 16 years to pension age £213.59 2018 MIS rate for a single working age adult Pension age £195.90 2018 MIS rate for a single pensioner

The JRF MIS includes a standard estimate for housing costs. It is generally accepted that due to the complexity and variable costs of current housing provision, it is problematic to include housing costs within a CBI.44 On the basis that a CBI should be a regular, consistent amount across the population and not subject to means-testing (which currently applies to benefits that cover housing costs), the payment levels in Table 7 exclude costs for housing. They also exclude the costs of childcare. It is anticipated that current benefits and support relating to housing and childcare (as well as other specific benefit needs) are retained alongside a pilot CBI. The detail of retained benefits is described in Section 5.5.

For the Low-Level CBI Payment the group agreed that a CBI set at less than people currently receive through benefits would not be ethical. Therefore, the low-level CBI payment is based on the current level of social security payments which are common at different ages (child tax credits, unemployment benefits and new state pension). The low-level CBI payments (Table 8) broadly reflect 2019 rates of current out-of-work benefits for different age groups. The proposed payment levels do not include additional entitlements added to means-tested benefits for some carers, those with limited work capability and disabled people (including disabled children). Like the higher rate of CBI, the low rate also does not include amounts for housing or childcare costs.

A CBI at a similar level to the entitlements which it replaces provides an opportunity to test the effect of an unconditional income with no change in the level of income. The higher level of CBI provides an opportunity to test the effect of both an increased income and a move to unconditional payments.

xix For more information on Minimum Income Standard, see https://www.jrf.org.uk/report/minimum- income-standard-uk-2018 xx Davis A., Hirsch D., Padley M. and Shepherd C. (2018) A Minimum Income Standard for the UK 2008- 2018: continuity and change, Joseph Rowntree Foundation, York, pp. 1

58 Table 8: Proposed Low Level CBI Payment

Age Range Payment Rate (per week) Basis for CBI Rate 0 to 15 years £84.54 (payment to main Rate of child tax credit family rate and first carer / parent) child rate (£63.84) plus Child Benefit eldest child rate (£20.70) 16 to 19 years £84.54 Reflecting rate of 16-19 year olds who are still in approved education: Rate of child tax credit family rate & 1st child rate (£63.84) plus Child Benefit eldest child rate (£20.70) 20 to 24 years £57.90 Rate of jobseeker’s personal allowance for a single person aged 16-24 years 25 years to pension age £73.10 Rate of jobseeker’s personal allowance for a single person aged 25+ years Pension age £168.60 Rate equivalent to new state pension

The payment levels identified above are costed at 2019/20 levels and would require to be uplifted in any future use. However, there are complications associated with uplifting the rates. For the lower level, which is designed to be similar to the payment levels of the current social security system, the UK government’s plan for 2020/21 is for these to increase by the rate of inflation. The exception to this is that the state pension increase will be higher than inflation because of the “triple lock” policy. For the higher level, which is linked to the MIS, account would need to be taken of the costs of living on low income and even a standard inflation uplift may not adequately adjust for this. The payment levels will therefore require to be revised for any future pilots based on these factors.

5.5 Interaction with Social Security

A key principle of the CBI feasibility study is that pilot participants (particularly vulnerable and low- income groups) are not financially worse-off either during or after the pilot. This is difficult to achieve without retaining some support to meet the greater financial needs of some groups. Therefore, the Steering Group explored a model which suspends participant access to some existing entitlements for the duration of the study, but retains benefits relating to additional needs associated with disability, work capability, housing and childcare support (see Section 8.2 for further detail).

For the avoidance of doubt, it should be noted that for benefits retained alongside a CBI, some conditions associated with eligibility will inevitably remain. For example, entitlement for housing, childcare and disability support is currently based upon individuals meeting certain eligibility conditions. While it is not within the remit of the Steering Group to redesign welfare support associated with these entitlements, it is strongly suggested that eligibility assessment is based on the principles of fairness, dignity and respect.

A list of the suggested suspended entitlements is outlined in Table 9. The suspended entitlements are proposed to be the same for both levels of CBI. The inclusion of pensioners is in line with the current approach of adhering to the universal principle of CBI and desire to have a saturation site within the pilot.

It is proposed that the low rate CBI for those of pension age is set at the same level as the new

59 state pension (£168.60). The Steering Group recognise the likely complexity of CBI interaction with the variations of state pension entitlements and that pensioner participants will have a range of existing payments, premiums and top-ups. To help avoid detriment, we want to ensure participants receiving payments in excess of the new state pension rate continue to have access to this amount and those on lower incomes continue to get any passported entitlements.

There are a number of significant legislative challenges and risks associated with these proposals, in particular with delivering a CBI which suspends access to a number of entitlements. Ensuring participants are not at risk of detriment by retaining access to some benefits, premiums, additions and elements will require detailed unpicking of the current social security provision. This requires changes to a range of Acts and regulations which are mainly the responsibility of the Department for Work and Pensions and Her Majesty’s Revenue and Customs. This will be complex and time- consuming.

The feasibility of this model in the context of the current social security system is explored further in Section 8.2.

Table 9: Suggested Benefit Entitlements to be Suspended During CBI Pilot Study

Benefit Entitlements Suspended for Pilot Durationxxi • Income Support (Personal Allowance) • Income-based Jobseekers Allowance (Personal Allowance) • Income-related Employment and Support Allowance (Personal allowance) • Child Tax Credit (Family Element plus Child Element) • State Pension • Child Benefit • Carer’s Allowance (Basic Rate and Scottish Supplement) • Universal Credit: Standard allowance for Single person • Universal Credit: First child / subsequent child payments

Interaction between a CBI pilot and Scottish Government’s new Scottish Child Payment has not been included within this feasibility report due to the recent announcement of the policy.xxii It is possible (and likely) that following submission of this feasibility report, there will be more developments within Scottish and UK social security which will have implications for a pilot CBI. To fully consider the implications of benefit suspension and legislate to mitigate any adverse consequences, it will be necessary to undertake substantial and further expert analysis on any proposed model of benefit interaction for the proposed pilot years (particularly within a changing social security landscape).

The treatment of a CBI payment for the calculation of the retained means-tested benefits that are not suspended will have an influence on how much individuals would receive from these benefits. xxi This table is set out in the context of the CBI characteristic of payments being made to the individual. It is not the intention that any couple rates would continue to be paid within these existing benefits. Instead, within this model, a couple would each receive the CBI payment. xxii For further details see: https://www.gov.scot/publications/scottish-child-payment-factsheet/

60 We considered the following options for the treatment of CBI: a) CBI (both payment levels) is counted as income; b) CBI (both payment levels) is disregarded as income; c) A hybrid approach, where: • Low level CBI is counted as income; • Within the high-level CBI: A value equivalent to the low level is counted as income, but the remainder of the CBI payment up to the high level is disregarded as income.

A CBI which is disregarded for benefit calculation would have no impact on the amount of money received from means-tested benefits. However, a CBI that is counted as income would reduce the value of the retained means-tested benefit and by consequence, could result in the loss of passported benefits. One example of this, that could clearly be a work disincentive, is the potential that access to childcare costs currently available would be ended. A parent or responsible carer in receipt of universal credit can access up to 85 percent of their childcare costs. This is capped at £646.35 per month for one child and £1108.04 for two-plus children. If the CBI is counted as income it could end the universal credit award and any associated support with childcare costs. In this scenario even meeting the childcare costs for one child would not be compensated by a low level CBI (only equivalent to around half the childcare costs) and take up the majority of a CBI paid at the higher level. This is just one of the potential detriments associated with ending passporting within the benefit system that a CBI pilot would have to interact with.

If the CBI payment was disregarded as income for the calculation of means-tested benefits, this would reduce the risk of detriment through the loss of benefit entitlement for participants and through the loss of passported benefits.

5.6 Interaction with Tax System

The UK tax and benefit systems are closely linked despite being administered by different organisations. With regards to a CBI pilot or policy, it is essential that tax and benefit interactions are considered alongside each another.

Income tax is the only currently used tax that could raise the required revenue for financing a CBI on its own.28 There is little evidence on the potential impact and sustainability of other sources of taxation for this purpose. Financing a CBI from income tax provides an opportunity to restructure the tax system to a more progressive form, helping to address poverty and inequality. As part of economic modelling of a national CBI policy in Scotland, Section 9 provides details of income tax changes required to fund a national CBI at the specified levels.

In order to provide relevant and sound evidence on the behavioural and economic effects of a national policy, a CBI pilot should mimic the national policy model as closely as possible. As such, this would require the CBI pilot to interact with the tax system. A pilot which was not able to interact with tax will have more limited learning for the behavioural, economic and social effects of a CBI policy. We suggest that a CBI payment should be included in the calculation of income for tax purposes. This would mean that all taxable income (CBI and non-CBI) which is above the

61 Personal Income Tax Allowance threshold (currently £12,500) would be taxed.xxiii The effect of this is to make the policy more affordable. The institutional feasibility of this position in relation to benefit interaction is considered in detail in Section 8.2.

5.7 Pilot Duration

As part of the evaluability assessment process we identified a range of intended outcomes that a CBI pilot might achieve (see Section 6). As part of this process, we considered how many years a pilot would need to run to be able to assess whether these outcomes were realised. It was noted that the time required to see effects varies between outcomes, with some taking much longer to work through (e.g. reduced income inequality and improved population level health and wellbeing).

The group also considered the duration of other contemporary CBI pilot studies underway or in progress around the world during general discussions with basic income experts at the Basic Income Earth Network (BIEN) Congress in August 2018.xxiv Experiments in the Netherlands, Finland and Canada are proposed to be between two and three years in duration. Reflecting on these discussions in our International Learning Report, it is suggested that, “a 2 or 3 year pilot is sufficient to measure outcomes, however this requires patience for results and an acknowledgement that time spent on a pilot needs to be respected”.14

Based on this analysis the Steering Group proposes a pilot duration of three years with a one-year preparatory period in addition to this. A three-year pilot would provide sufficient time for participants to adapt to a CBI and for shorter-term outcomes to emerge. A pilot longer than three years may be at risk of weakened political confidence and commitment if there was a change in government, with resulting premature ending of pilots and a loss of learning from the process.

A one-year preparatory period is important for avoiding delays which could curtail the duration of the pilot. A discussion in February 2019 between Steering Group members and the co-principal investigator of the evaluation team for the cancelled pilot in Canada reported extreme time pressures prior to the commencement of the study. These pressures had knock-on effects in terms of decision-making and constraints on the pilot. It would be possible to avoid such time pressures by incorporating a preparatory period. This would allow time for recruiting and preparing participants, trouble-shooting issues and gathering baseline data, therefore ensuring an effective and robust pilot is ready to start on the first day of the three-year pilot period.

5.8 Pilot Populations

There are two broad approaches to piloting a CBI: a saturation approach where a CBI is given to everyone (of all ages and income levels) within a defined geography; or a targeted approach in which everyone in a group of interest in the wider population receives a CBI e.g. care leavers, people who are disabled or unemployed, people from black and minority ethnic communities, disabled people or people who are unemployed.

A saturation approach reflects the CBI principle of universality (where the ‘universe’ includes all of those in a defined geographic area). An historical example of a saturation experiment is the town xxiii In the commissioned economic modelling work, it is assumed that the Personal Income Tax Allowance threshold is removed which contributes to the revenue needed to fund the proposed CBI. However, the proposal is that it should not be removed in the pilot study. xxiv The Carnegie UK Trust funded a delegation from the Steering Group to take part in an international study visit to the 2018 BIEN Congress in Tampere, Finland.

62 of Dauphin, Manitoba, Canada in the 1970s45 (although the CBI was only given to residents whose income fell below a certain level as it was a negative income tax style intervention). Experiments such as those in Kenyaxxv and the small, village-wide pilot in Ugandaxxvi are more contemporary examples of saturation type experiments.

It is theorised that the universal nature of CBI may lead to a more united, inclusive and harmonious society and a reduction in stigma associated with targeted benefits.28 Research exploring stigma associated with claiming benefits in contemporary Britain, concluded that benefit stigma is primarily driven by the perception that claimants are ‘undeserving’.46 The study reported that stigma plays a role in the non-take-up of benefits or delaying of claims and has a clear negative impact on claimants’ feelings of self-worth. It is argued that stigma may be avoided if there are feelings of solidarity between recipients and if benefits are viewed as entitlements, borne from prior contributions, or based on citizenship. In this respect, a CBI could be regarded as a mechanism for developing solidarity between all individuals due to its universal approach to income entitlement.

A saturation type experiment where all residents within a geographical area are in receipt of CBI would allow this hypothesis to be tested. It also provides the opportunity to assess individual, household and community-level impacts associated with universal receipt of a CBI.

Community-level impacts are an important aspect of the group’s theory for how a CBI might impact on the outcomes of interest. The hypothesis is that as a result of everyone in an area receiving the CBI, there will be community-level impacts (also known as ‘social multipliers’) over and above those which might occur through the direct, individual impacts of receiving the CBI. Such community impacts include volunteering, informal caring networks, and the creation of new social enterprises, businesses and clubs.

There are some examples where community-level impacts have been detected as an outcome of an unconditional-type payment,10 ,11 however the evidence is sparse with few pilot evaluations designed to specifically capture such impacts. To test for community effects, the Steering Group are interested in designing a pilot model and evaluation plan which provides maximum opportunity for them to occur. We propose that a saturation study provides an opportunity to measure and evaluate the social, economic and community effects of a universal payment in a way which has never been undertaken before and would provide valuable evidence for the future design of social security.

We considered an alternative, targeted approach, but rejected it on the grounds that it would not meet the universal principle of a CBI and would not generate community-level outcomes associated with a saturation sample. The Steering Group also considered a pilot in which the CBI would be given to a random sample of the general populationxxvii but this was also rejected on the grounds that it would not generate community-level effects because CBI recipients would be spread across a large geographical area.

Following consideration of the different options, our recommendation for any pilot would therefore be a saturation approach. Further detail on the size and number of communities required to evaluate a saturation approach in a way that enables us to explore community-level effects and different effects between different groups within the population at a reasonable cost are discussed further in Section 6. xxv Details of the study in Kenya can be found here: https://www.givedirectly.org/ubi-study/ xxvi Details of the study in Uganda can be found here: http://eight.world/ xxvii A random sample could be stratified to ensure that it reflects the composition of the population of interest in terms of characteristics such as age, ethnicity, prevalence of disability, socio-economic status etc.

63 5.9 Pilot Model Payment Options

There are a number of methods which could be used to transfer CBI payments to pilot participants. Contemporary international pilot studies demonstrate various methods including pre-paid debit cards, direct credit transfers and secure mobile phone applications. Some of these also incorporate the use of social, local or digital currencies to measure effects on the local economy (e.g. the experiments in South Korea, Barcelona and Brazil).

Not all methods will be appropriate for all pilot studies; the suitability varies according to the pilot design requirements and outcomes of interest.

We are committed to designing a pilot model which tests as many of the principal characteristics of a basic income as possible. Suitable payment methods therefore need to be capable of meeting these characteristics, as well as other criteria which ensure a reliable, cost-effective mechanism which satisfies legal obligations for participants.

A list of the necessary criteria for a suitable payment method is outlined below:

1. Reliable, regular, individual An effective, reliable and accurate method of delivering a regular (weekly, fortnightly, monthly) CBI payment individually to participants for the duration of the pilot (recommended to be three years).

2. Appropriate for saturation population The method must be suitable for all population groups including people who are in receipt of benefits, those who do not receive benefits, and homeless people or traveller communities who may not have a fixed registered address. It must also be flexible to accommodate changing CBI payment rates if participants move up to a new age bracket during the study. It should also allow the addition of new participants (through births or inward migration) and removal of participants (due to death or outward migration).

3. Cost-effective administration The method should be cost-effective and provide good value for money throughout the duration of the pilot. This includes during set-up of the facility, administration and ongoing management. A customer service facility should also be provided to ensure payment problems and queries can be addressed.

4. Satisfies legal requirements It is essential that any method satisfies legal obligations and requirements such as protecting the identity and data of participants and is compliant with GDPR regulations.

5. No restriction on use The payment method should not restrict how the participant chooses to use the payment unless this is an explicit design feature of the study. For example, a current experiment in Barcelona distributes 25% of basic income payment via a social digital currency called ‘rec’, with the effects of this local currency being evaluated as part of the experiment. We do not recommend any restrictions on how the CBI can be used.

Table 10 describes a list of options for payment methods within a pilot study together with an example of where they have been used and an assessment of their opportunities and challenges.

64 Table 10: Options for CBI Payment Method

Payment Method Example where used Opportunity Challenge Direct credit Finland Existing accounts used, Existing accounts may be transfer to existing don’t require custom joint and not individual bank / building design or development which could interfere with society account in costs. potential behavioural effects the name of the Same method used of an individual CBI recipient by DWP so many Would require bank account recipients will be details for all participants. familiar with this method and should not interfere with Some participants may their ‘normal’ way of not have bank accounts accessing and using (e.g. travellers or homeless funds. people). Prepaid debit card Stockton Economic Participants should Debit card would need to (procured by pilot Empowerment be able to use card in be procured by pilot project project group) Demonstration their preferred way e.g. team, increasing project (SEED), Stockton, shop/online purchases costs. California and extract cash. Require a secure process for Barcelona Not linked to a crediting the prepaid card. traditional bank Need to ensure appropriate account so useful for data and identity security participants who do compliance. not have one. Secure Mobile App Uganda May be useful for An App may have to be participants who do custom built to required not have a traditional specification which could be bank account. costly. Depending on security May require providing smart permissions, may phones to participants who be possible to track do not have a compatible spending use. device which could be costly. Some participants may not trust a mobile banking app and therefore could restrict their behaviour. Need to ensure appropriate data and identity security compliance.

65 Payment Method Example where used Opportunity Challenge Custom CBI Y-Combinator Useful for participants Service would need to be Registered Bank Research,xxviii who do not have procured which will add Account (online Oakland, California a traditional bank cost. current account and (e.g. Chime Banking) account. Need to ensure appropriate debit card) Participants should data and identity security be able to use card in compliance. their preferred way e.g. Some participants may not shop/online purchases have access to internet and extract cash. services to access an online account. A mobile device may have to be provided which will add cost.

Unless a specific feature of a payment method is to be explicitly tested and evaluated (e.g. a digital currency via a secure mobile app), care must be taken to avoid using a payment method which interferes with participants’ ‘normal’ ways of accessing and using funds. To do so may inadvertently result in the addition of another intervention which may complicate the behavioural effects associated with a CBI pilot.

Our interest is not to test the effects of a local or digital currency, therefore the preferred method will use pounds sterling as the form of currency.

The most cost-effective and secure method of paying CBI to participants would be the first option: direct credit transfer to an existing bank or building society account. This is the preferred method of DWP, HMRC and Social Security Scotland for delivering benefit payments. Current benefit entitlements from DWP and HMRC can be paid into a bank, building society or credit union account. Payments to Post Office card accounts are currently being phased out and will no longer be available after November 2021.

Although this option is preferable, it would still require mitigation to address the challenges associated with the method. It would need to be coordinated with the agreed delivery organisation to ensure the secure collection and storage of participant account details, with a facility to allow participants to update details when necessary.

An individual with no fixed address can claim benefits as long as they can provide an alternative address. This could be the address of a friend or family member, a hostel, job centre or day centre. This is not the same as a residency test, the requirements of which would still have to be met by the individual.

In the case where pilot participants do not have an existing bank account or cannot access one due to having no fixed address, it may be possible to adopt the same methods currently used by DWP and Social Security Scotland. In this situation it is advised that individuals open a relevant account for payment if they do not already have one. If the individual cannot open or manage a bank, building society or credit union account, they may be able to use the Payment Exception Service to collect their money. The Payment Exception Service is delivered using i-movo, a secure digital xxviii Y Combinator Research is an American research company currently designing an experiment to explore the impact of basic income in the United States.

66 voucher system. It enables vouchers to be distributed by SMS, email, direct mail or mag-stripe card and are accepted across various sectors including Banking, Consumer Goods, Government, Newspaper Publishing, Retail & Utilities and is registered with the Financial Conduct Authority (FCA) as a Small Payments Institution. For devolved benefits, Social Security Scotland currently uses these existing arrangements, however it may consider developing a longer-term, user-led alternative payments service.

67 Section 6: Evaluability Assessment

An evaluability assessment is a way of working through whether and how a policy or an intervention such as a CBI can be effectively evaluated. It involves clarifying with stakeholders the intended and unintended outcomes of the policy and assessing whether and how these can be measured with the time and resources available.

Section 8.2 documents the institutional barriers to the implementation of a CBI pilot of the kind proposed in Section 5. These barriers mean substantial uncertainty remains about the form any pilot CBI might take or the timescales over which it might be introduced. This has implications for the degree of certainty with which it is possible to assess the best way to evaluate the proposed pilot. For example, barriers to putting in place all the tax and benefit changes proposed in Section 5 may mean that different pilot models need to be considered. This would affect the administrative costs of the pilot, but without clarity on the scope of the pilot that could and would be carried out, the actual administrative costs cannot be estimated with any degree of certainty.

Likewise, if a pilot CBI at the levels and with the coverage proposed in Section 5 is not possible, this will have implications for the effects it is reasonable to expect, the evaluation questions we might want to ask in relation to those effects, and the measures used to assess them. It would also affect the size of the sample needed to measure them (see Section 6.3.2 below). In addition, the barriers identified are likely to mean a substantial lead in time is required to prepare for the pilot and the evaluation. During this time, the relevant stakeholders, their policy concerns and therefore the questions they would want a pilot to answer might all change.

Therefore, the evaluability assessment does not arrive at a detailed plan for the evaluation. Rather, it discusses the advantages and disadvantages of different potential evaluation designs and highlights the key features of the design we think would be most appropriate. It then discusses the practical issues that would need to be addressed in implementing the proposed design in terms of defining and measuring the outcomes of interest, determining the size of the study sample, selecting the study areas, understanding the process by which any changes arising from the CBI come about, and estimating the potential cost of the pilot.

One fundamental question is whether a pilot study based on the model of CBI proposed in this report offers the best way of evaluating the potential impact of a CBI, were it to be rolled out across the country as a whole. It has been suggested that, in a relatively short pilot on a small scale (relative to Scotland as a whole), a CBI could not have the transformative effect that some people think it would have if it were rolled out permanently across Scotland. For example, long-term commitments to training or the establishment of new enterprises would not occur if the CBI is only available for the three years proposed in Section 5. In these circumstances, absence of intended effects may reflect the limitations of the approach to piloting, rather than the ineffectiveness of a CBI were it rolled out country-wide. The absence of a CBI in neighbouring areas might also affect the outcomes in the intervention site, for example, by affecting people’s decisions about where they live and work.

Perhaps most importantly, the institutional barriers discussed in Section 8 may mean it is not possible to pilot a CBI resembling the model proposed in Section 5 in terms of all the tax and/or benefit changes that would be required. As a result, the pilot may give a false impression of the labour market responses that are central to many of the economic and distributional effects of a CBI.

68 In a pilot scenario there are potential impacts that would not be seen in a rollout of CBI. For example, there may be effects on migration in or out of the area to take up employment. Potential ways of mitigating some of these are discussed in Section 6.3.2.

Whilst these limitations are important, it is also important to consider the limitations of alternative approaches to piloting a CBI, such as smaller scale pilots or modelling approaches. These are addressed earlier in the report.

Therefore, while there are limitations in terms of the impacts we would expect a pilot study to have, a pilot on the scale and design proposed and an accompanying evaluation might still generate useful learning to inform any decision about whether to introduce CBI across Scotland. Even with the limitations described earlier, the CBI pilot proposed would represent a significant change in the choices people face, with the potential for substantial impacts, positive and negative. Conversely, having no empirical evidence on the impact of a CBI, would itself be a major limitation if and when government were making a decision about whether to roll out a CBI more widely. The balance of these arguments needs to be considered in deciding whether to run a pilot on the scale proposed, given the very substantial potential cost of the policy and the importance of the outcomes it aims to achieve.

The remainder of Section 6 describes the process we have undertaken to identify a robust approach to evaluating a potential CBI pilot. The process involved four workshops in which the Steering Group:

• Developed a draft theory of change for a CBI in Scotland (Section 6.1). The theory of change maps out both the outcomes it was hoped to see and the possible unintended outcomes that could result from a CBI in Scotland. • Prioritised these outcomes with a view to focusing on those that are more likely to be measurable across a sample of the population in the course of a three year pilot (Section 6.2) • Identified the preferred approach to the evaluation based on the proposed model of CBI (Sections 6.3 and 6.4) • Considered the ethical issues that an evaluation of a pilot CBI might raise (Section 7).

6.1 Theory of Change

A theory of change “explains how activities are understood to contribute to a series of results that produce the final intended impacts.”xxix It includes both the intended outcomes of a policy or intervention and the potential unintended consequences, both positive and negative. It highlights some of the external factors that might affect the outcomes of interest. It also identifies the short- and medium-term outcomes that would need to be achieved if the longer-term intentions of the policy are to be realised.

Figure 3 presents the proposed CBI theory of change developed over the course of the workshops and subsequent discussions within the Steering Group, based on available evidence. Columns 2 and 3 describe the potential intermediate and longer-term population level outcomes that might

xxix Rogers, P. (2014). Theory of Change, Methodological Briefs: Impact Evaluation 2, UNICEF Office of Research, Florence. https://www.betterevaluation.org/en/resources/guide/theory_of_change

69 follow from a CBI, if rolled out across the population over a longer period.xxx

As noted earlier, in a time-limited pilot involving a sample population, it would not be feasible to assess whether the longer-term, population-wide, outcomes are achieved. Therefore, column 1 indicates the short-term outcomes that a CBI might achieve at individual and household level over a three year pilot study, which might be measurable in a sample of the population and which are plausibly linked to the intermediate and longer-term outcomes. For example, increased individual opportunities to make choices about informal caring, volunteering, or community activism might contribute to a community’s sense of being connected and empowered to influence decisions that affect the whole community. A CBI might impact on the local economy in the short-term by increasing entrepreneurial activity, and/or it might promote inclusive growth by facilitating training amongst lower-skilled, lower-paid workers and encouraging search for higher quality jobs.

Some of the outcomes we identified are broad but could be measured using a range of indicators. The outcome ‘increased opportunities to make life choices’, for example, could be measured using data relating to volunteering, informal care, involvement in community activity, labour market and leisure activities. Section 6.2 below indicates some of the potential measures that could be used.

We outlined some of the assumptions underpinning the theory of change, particularly in relation to institutional and psychological feasibility (see Sections 8.2 and 8.3), which could be tested in the course of an evaluation. We also considered some of the potential unintended consequences (positive and negative) of a CBI pilot as well as the external factors that may influence whether and what is achieved, even within the context of a pilot. These are summarised in Appendix 3.

The CBI theory of change is not fixed and we would recommend that it should be developed over time as new evidence becomes available, new stakeholders become engaged in the discussion around CBI and/or as stakeholders identify new areas of interest. However, the number of outcomes that can be included in an evaluation is constrained, first, by the likelihood that change will happen during a (relatively short) pilot; and second, by the feasibility and costs of collecting (additional) data. This is discussed further in Section 6.2.

xxx These population level outcomes also feed into several of the outcomes in the Scottish Government’s national performance framework (Figure 3, column 4). https://nationalperformance.gov.scot/national- outcomes (accessed 24th June 2019)

70 Figure 3 – Summary Theory of Change

SHORT TERM OUTCOMES INTERMEDIATE OUTCOMES LONGER TERM OUTCOMES RELEVANT NATIONAL (2-3 YEARS/PILOT PERIOD) (IF CBI ROLLED OUT) (IF CBI ROLLED OUT) PERFORMANCE FRAMEWORK OUTCOME

Reduced income insecurity Sustained reduction in levels Reduced income inequality People tackle poverty by sharing of destitution opportunities, wealth and power Increased income Reduction in educational more equally Reduction in absolute and relative attainment gap Decreased use of emergency support poverty rates, including child poverty People grow up loved, safe and Improved population health respected so that they realise Reduced debt Improved educational attainment and wellbeing their full potential Reduced poverty, including Increased sense of community Reduction in population level child poverty People are healthy and active ‘control over destiny’ crime rates People are creative and their Increased opportunities to make Increase in proportion of population More ecient/simpler social security vibrance and diverse cultures are life choices choosing informal caring role system: reduction in social security expressed and enjoyed widely Improved health and wellbeing administration costs Improved social connectedness People live in communities that are Contribution to inclusive/economic inclusive, empowered, resilient Improved experience of social Reduction in local crime rates growth and safe security services - reduced stigma (including domestic violence) Reduced barriers to labour market Respect, protect and fulfil human Welfare policy informed by learning/ participation rights and live free from evidence from CBI discrimination Increased uptake of opportunities for acquiring training, skills, People are well educated, skilled qualifications and able to contribute to society

Increased entrepreneurial activity Scotland has a globally competitive, entrepreneurial, inclusive and Improved labour market participation sustainable economy (in fair work) People have thriving and innovative businesses, with quality jobs and fair work for everyone

71 6.2 Outcome measurement

As noted above, the evaluation would focus on measuring the impacts on those outcomes identified in the theory of change with the potential to change within the three-year period of a proposed pilot. In designing an evaluation, a distinction is often made between ‘primary’ outcomes and ‘secondary’ outcomes.xxxi The primary outcomes are those for which it is most important to be able to detect the change of interest. In this study, based on the discussion with the four council areas and the Scottish Government, we recommend that the primary outcomes should be reduction in child poverty, reduction in poverty and reduction in unemployment. The other ‘secondary’ outcomes in the theory of change would also be measured, either quantitatively or qualitatively, as part of the evaluation, but the study design would be based primarily on ensuring that the evaluation were able to detect change in the primary outcomes (see Section 6.3 below).

There are a number of potential sources of data for measuring individual/household and community outcomes, and the process of implementation and delivery. These include:

• ‘Universal’ administrative data routinely collected by, for example, HMRC, DWP and local authorities • Administrative data routinely collected as part of the CBI delivery system created to administer the CBI pilot, if it goes ahead • Existing population surveys (Scottish Household Survey, Scottish Health Survey, Labour Force Survey, Scottish Attitudes Survey etc.) • ‘Bespoke’ survey data • ‘Bespoke’ qualitative data e.g. through in-depth case studies.

Appendix 4 groups the short-term outcomes in the theory of change into five domains. It includes a domain covering background socio-demographic variables and one to capture the processes of delivering a CBI, and the mechanisms underpinning any observed outcomes. Based on a review of existing surveys, it also summarises the different potential data collection sources. Some of the data sources could be used to collect data in more than one ‘domain’. For example, a survey (whether self-administered or administered via an interview (analogous to the Scottish Household Survey or Scottish Health Survey)) could collect socio-demographic, income, employment/economic status and wellbeing data, etc. In addition, in-depth qualitative case studies could capture more experiential data, for example, on health and wellbeing, individual and community empowerment, changes in experience of the social security system following introduction of a pilot CBI, and/or any challenges around implementation.

There are limitations in the routine data available. In relation to the primary outcomes, for example, poverty is estimated using the data from the Family Resources Survey and data on benefit uptake. Coverage of these sources in the ‘saturation’ sample in one or a small number of pilot sites would be inadequate to detect changes in poverty resulting from the CBI. A survey would therefore be required in which the whole sample were invited to provide data.

Routinely available unemployment data at a local level are based on the claimant count version of the indicator, which understates true unemployment because many unemployed people do not xxxi Andrade, C. 2015, The primary outcome measure and its importance in clinical trials, J Clin Psychiatry. 2015 Oct;76(10):e1320-3. doi: 10.4088/JCP.15f10377 (accessed 23 July 2019); http://www.consort- statement.org/resources/glossary#S (accessed 24th June 2019)

72 claim benefits.47 The more valid measure in research of this kind is the International Labour Office (ILO) measure which includes all those available for and actively seeking employment, regardless of whether they are claiming benefit. These data are collected through routine surveys such as the Labour Force Survey. Again, coverage would be inadequate to detect changes in unemployment within the pilot sites, so would need to be collected by a bespoke survey of the sample population.

The extent to which a pilot could influence data collected routinely or via the administration of a CBI, would be contingent on whether and how the relevant agencies adapt existing systems to accommodate a CBI. If a separate CBI delivery agent were established, then this would provide an opportunity to influence the data collected from participants (contingent on appropriate information governance procedures).

Although routine surveys will not sample a sufficient number of people in the intervention populations to be useful, and would not ask the full range of questions that are relevant to this evaluation, there is the potential to use routine administrative data as an adjunct to the survey data we propose to collect. Routine administrative data (such as hospital admissions, benefits claims, housing applications) are held for the entire population. It is possible to anonymise such data to make it suitable for research purposes, and to then compare this between people living in intervention and comparison areas (for example, using a synthetic control which finds a suitable comparison group from amongst the whole population to compare with the population given the CBI). This is a relatively cheap evaluation approach that could be planned late in any CBI pilot process. The main limitations of this approach is that it contains a very limited range of relevant data (often pertaining to the use of services such as the NHS, births, deaths, marriages and educational achievements) and there can be systematic data problems such as people being registered at incorrect addresses. These limitations notwithstanding, if a pilot were to go ahead this would be a very useful additional quantitative element in any evaluation.

Whatever use is made of routine data, ‘bespoke’ survey and qualitative data are likely to be required. A survey could focus on ‘core’ outcomes of interest, specifically relating to income, labour market activity and health and wellbeing, with other outcomes being explored via other methods or with sub-groups rather than the whole sample and control group, taking into account the impact on statistical power and generalisability. For example, different issues could be explored using ’modules’ asked at different times (analogous to the Scottish Social Attitudes Survey). Contingent on receiving the relevant permissions, questions contained in pre-existing surveys such as the Scottish Household Survey, Scottish Health Survey, Labour Force Survey, Family Resource Survey could form the basis of the research instruments if a pilot were to progress. This would help ensure the validity of the research tools, while also allowing for a degree of comparability with population- wide data. Consideration could also be given to the potential for ‘piggy-backing’ on existing surveys, contingent on being able to extend the reach to the CBI populations. These surveys gather demographic information that would be equally relevant to the CBI pilot, but their primary focus would be information not directly relevant to the CBI, so either interviews would need to be lengthened or follow up interviews undertaken, both of which would add to the burden on respondents as well as eroding some the cost savings from piggy-backing on existing surveys. This would need further discussion with the relevant partners in the Scottish Government Chief Statistician’s office.

Appendix 4 indicates that it is possible to identify appropriate survey instruments and data sources for many of the outcomes of interest. However, detailed design work would need to be undertaken following any decision to progress with a pilot. This would need to reflect the outcomes of interest at that stage, and the agreed model of pilot pursued. It would also need to balance the number

73 of outcomes measured with what it is practical to measure within the time frame. This will be influenced by the relative balance between data that would need to be collected from participants, whether through surveys, interviews, focus groups etc., and administrative and routine data that can be collected and analysed ‘remotely’. As noted above, these are often limited in terms of the number of outcomes they measure and the scope to disaggregate them to explore differences by sociodemographic characteristics. On the other hand, placing too many demands on participants through the use of extensive surveys might generate ‘research fatigue’, which has implications for non-response rates and data completeness. It also has implications for the potential cost of the evaluation (see section 6.5).

6.3 Study design

Studies based solely on outcomes observed in an area where a policy or service is introduced are prone to confounding. That is, the change in outcomes observed may be attributable (partly or wholly) to secular trends in the variables of interest and the factors driving those over time. Therefore, to robustly assess the impact of a CBI on the outcomes of interest it is necessary to have both an ‘intervention’ group or groups who receive a CBI over the pilot period, and a ‘control’ group who do not. The control population must be comparable to the population receiving the CBI to increase the likelihood that any differences in outcomes are attributable to the CBI, rather than the differences between the control and intervention groups.

We recommend that for this study, the most appropriate and robust evaluation design would be an outcome evaluation comprising a Randomised Control Trial (RCT) with outcomes measured at baseline and at least one wave of follow up in areas receiving the CBI and a control group that do not. Further waves would be desirable but would face considerable cost and logistical issues – see Section 6.6. The intervention groups should each comprise one or a small number of communities selected at random from a list of communities based on criteria discussed further below. Everyone in the intervention community or communities would get the CBI, in line with the ‘saturation approach’ suggested in Section 5.

Quantitative data collected through surveys or available in routine sources would be used to measure the changes in outcomes in the intervention groups compared to changes in the comparison population. This form of evaluation is termed a ‘difference in differences’ approach whereby the impact of CBI would be measured as the difference in the change in the outcome measures between the intervention and comparison populations.

The possible measures for this were discussed in Section 6.2 above and the potential costs of survey data collection and analysis are discussed in Section 6.5 below.

Within the pilot community/communities, everyone receiving the CBI would be invited to take part in the evaluation. The alternative would be to draw a sample from a wider population receiving the CBI. However, this would make the pilot prohibitively expensive because it would increase the number of people that would need to receive the CBI to get a big enough sample for the evaluation to be adequately powered.xxxii xxxii For example, the power calculations below suggest we need around 800 people to be powered to detect differences in the primary outcomes between the intervention and control groups for the high level of CBI, 1600 if we want to explore the impacts of the CBI on men and women separately. If a 10% sample of the group receiving the CBI were drawn, this group would need to comprise 16,000 people to achieve the required level of statistical power. At the low level of CBI, we need 14,600 people to be adequately powered, or 146,000 receiving the CBI if a 10% sample were drawn.

74 The Steering Group also considered a pilot in which the CBI would be given to a random sample of the general population stratified to ensure that it reflects the composition of the population of interest in terms of characteristics such as age, ethnicity, prevalence of disability, socio-economic status etc. However, a random sample drawn from the general population would mean that very few people from the same community would receive a CBI, limiting the potential economic and social impacts that might arise across the community as a result of the whole community receiving it (for example, the creation of a childcare co-operative, a new volunteer-led initiative, etc.).

Concern has been expressed about the appropriateness of an RCT design for an evaluation of this kind. Realist evaluation stresses the importance of understanding the mechanisms or processes by which an intervention might or might not work, for example, whether and how an intervention is implemented, whether there are teething problems in its delivery, or whether aspects of the intervention are ill-suited to the needs of particular population groups.48

One such ‘mechanism’ is the context in which a complex community-based intervention is implemented. Context refers to the external influences on a community and the specific social and economic circumstances in a particular location that might determine the impact that an intervention has when introduced in that context. According to this argument, an RCT prevents a full exploration of the effect of context on outcomes by defining a study population of ‘comparable’ people or communities and then assuming that the intervention area(s) and the controls picked at random are similar in every respect other than the intervention. This makes it difficult to explore the impact on outcomes of differences in context between areas.

However, guidance on the development and evaluation of complex interventions suggests that randomisation should always be considered to avoid the biases that can occur when the people or areas that receive an intervention differ systematically from those that do not, in ways likely to affect outcome.49 The guidance also stresses that outcome evaluation in the context of a RCT can and should be combined with a process evaluation to understand whether and how the intervention has been implemented, what the causal mechanisms might be and how contextual factors might influence the outcomes of interest.

We recommend this combined approach as the best way to evaluate a pilot CBI. The remainder of Section 6.3 explains how an RCT would be carried out. Section 6.4 describes the proposed process evaluation that would sit alongside the RCT.

6.3.1 Sampling strategy The sampling strategy needs to define:

1. Methods for ensuring that estimates of the impact of the intervention within the pilot community are robust. That is, a big enough sample is required to ensure that the estimates of impact are sufficiently precisexxxiii and a comparison group is required to avoid confounding. Both are required to be confident that any changes observed are likely to be the result of the intervention. Sample size is discussed in Section 6.3.

2. Methods for making the results generalizable. Generalisability requires that the evidence is generated from a context (or contexts) that is similar to the areas to which the evidence will be applied. This makes it more likely that the mechanisms underlying the effect will operate in a similar way outwith the pilot areas so that the outcomes are more likely to be similar to xxxiii That is, the samples need to be sufficiently ‘powered’.

75 those observed in the pilot area. The number and types of communities to include in the study are discussed in Section 6.3.3. The method for identifying the communities to be included is discussed in section 6.3.4.

6.3.2 Sample size Powering a study requires prior information on the size of the potential change in outcomes that the study needs to be able to detect. Trying to detect change in all the outcomes of interest may lead to very large sample sizes. This would increase the costs of both the CBI and the evaluation. Therefore, for the purposes of an evaluation, power calculations focus on the sample size required to detect change in the primary outcomes (defined in Section 6.2).

Although the proposed pilot study has not been powered to be able to detect change in the secondary outcomes, they would still be measured as part of the evaluation, either quantitatively or qualitatively. For outcomes measured in quantitative terms, information on the precision of the estimated effects would be provided. This would mean that although the evaluation would not necessarily be powered to detect changes in the secondary outcomes at conventional levels of statistical significance, these outcomes could still be measured. Differences between the intervention and comparison groups may still be found, if the effects are big enough in relation to the sample size.

In calculating the required sample, we have assumed:

• The prevalence of poverty and unemployment in the intervention and comparison populations are the same as Scotland overall.xxxiv • The CBI model at the low rate will be expected to have a smaller impact on relative poverty, such that we would expect, and want to be able to detect a change of two percentage points (i.e. from 26% to 24% for child poverty, and from 20% to 18% for poverty overall).xxxv • The CBI model at the higher rate will be expected to have a larger impact on poverty, such that we would expect, and want to detect, a halving of the poverty rate (i.e. from 26% to 13% for child poverty, and from 20% to 10% for poverty overall).

Existing evidence suggests that the impact on unemployment will not be large, but it is important to have adequate power to detect changes in this outcome. We have calculated the power to detect a three-percentage point change in unemployment. Unemployment rates have been as high as 8.7% in 2011 and 6% in the last two years, making a three-percentage point change a reasonable change to expect.

All of the assumptions above can be varied but in general, the smaller the size of effect we want to xxxiv This does not necessarily mean that study areas will be selected which have levels of these outcomes similar to Scotland as a whole. This is a working assumption for the initial calculation of the required sample size. The assumption can be varied to check the implications of choosing, for example, more deprived areas where levels of poverty and unemployment may be higher. The number and type of areas to include is discussed in Section 6.3.3. xxxv The latest data on the prevalence of poverty in Scotland are available from ‘Poverty and Inequality in Scotland 2015-16’. After housing costs, the prevalence of child poverty is 26% and for the population overall is 20%. Using the International Labour Organisation (ILO) definition of unemployment (which is not dependent on individuals claiming unemployment benefit), the latest unemployment rate for Scotland (from January 2019) is 3.3% (see https://www.ons.gov.uk/employmentandlabourmarket/ peoplenotinwork/unemployment/timeseries/ycnn/lms).

76 be able to detect, the bigger the sample required.

Table 11 provides the sample sizes required to detect changes in the primary outcomes in the intervention groups (assuming one group receives the high payment and one the low payment) and control groups. At the high rate of CBI, the required sample sizes are small (142 for child poverty, 196 for poverty overall and 792 for unemployment). At the low rate of CBI, the required sample sizes are much larger: 7,340 in both the intervention and control groups for child poverty and 6,494 in each group for poverty overall. The required sample size to detect change in unemployment is again 792 in each group because the effect size we want to be able to detect is the same, i.e. a three-percentage point change.

The evaluation would be designed to collect information on the impacts of CBI across all protected characteristics and socioeconomic position but we would not propose to power the study to be able to look at differential impacts for all of these groups, nor for intersectional impacts (i.e. for the sub-groups with multiple characteristics) because of the impact this would have on sample size and study cost.xxxvi We would still, though, measure outcomes in different groups, such as disabled people, and assess the sensitivity we have to detect any differences and indicate how certain we are about these.

We would want to understand the potential for differential impacts between men and women. With a saturation sample, approximately 50% of the population would be men and 50% women. This means we would need to double the sample size to understand the impacts of CBI separately for men and women.xxxvii Simply stratifying the samples by men and women would greatly reduce the required sample size but would underpower the study to detect different effects by gender. Such effects are of interest, given that it is likely there will be gendered differences in the impact of a CBI on outcomes such as labour market decisions, roles within the family and other caring roles. However, the ability to explore these comes at a high financial cost because of the additional sample size that would be required.

xxxvi The prevalence of disability in the population can be assessed in different ways. Using the definition from the Annual Population Survey (people who are between 16 and 64 years old self-report whether or not they have an illness which has lasted for more than 12 months which has kept them from working) it was estimated that in 2018 there were a total of 210,600 people with a disability (accounting for 4% of the 5,254,800 people estimated to be resident in Scotland in 2018). A saturation pilot approach in an area representative of Scotland overall would need to have the sample sizes detailed in Table 6.1 for people with disabilities. Given the prevalence of disability, this very substantially increases the overall population size required for sampling. At the low CBI rate this becomes prohibitively large (at 128,772 for child poverty, 113,930 for poverty overall, and 13,895 for unemployment; in the intervention and comparison groups). At the higher rate the required samples are more reasonable (2,491, 3,439 and 13,895 in the intervention and comparison groups for child poverty, overall poverty and unemployment respectively). The total population sizes required would be smaller if the intervention and comparison areas were to be more deprived and have a higher prevalence of disability. For example, the prevalence of disability in Glasgow is 5.7%, reducing the required sample sizes across the board by 30%. xxxvii i.e. comparing men in the intervention group with men in the control group, and women in the intervention with women in the control group.

77 Table 11: Required Sample Sizes by Primary Outcome and Level of CBI

Outcome Scotland average Percentage point Required sample prevalence change to be size in each group detected Low rate CBI Child poverty 26% 2% 7,340 Poverty 20% 2% 6,494 Unemployment 3.3% 3% 792 High rate Child poverty 26% 13% 142 CBI Poverty 20% 10% 196 Unemployment 3.3% 3% 792

Notes: 1. Power calculations assume that the intervention and comparison populations are equally sized, that we are aiming to achieve 80% power (i.e. 80% chance of detecting a real effect) and that we want to be able to detect both increases and decreases in our primary outcomes (i.e. a two-tailed test). 2. These sample sizes need to be doubled to look at effects in men and women separately.

6.3.3 Response rates, attrition and migration in and out of the pilot sites The sample sizes estimated above are the number of people we need to collect data on up until the end of the pilot to achieve the required level of statistical power. However, we would expect some people to decline to participate in the study, some people to decline to respond to the surveys and some attrition, i.e. loss of the initial population (including out-migration and deaths) over the duration of the pilot.

This needs to be factored into the evaluation design by increasing the size of the sample invited to take part in the study. However, the scope to learn from different studies about how big an increase in the sample size is required is limited by the fact that the degree of non-response for an intervention like CBI is likely to vary by context, by the groups involved, by the type of intervention and by the method of recruitment.

To inform the planning of the evaluation, we reviewed response rates in surveys and basic income studies elsewhere to assess the likely order of magnitude of these effects. For national surveys, response rates are typically between 50-65% of eligible participants and they have been falling steadily over time, which increases the risk of underestimating non-response if we base our calculations on those.

The limited information from other CBI experiments suggests that response rates are likely to be substantially lower than those in routine surveys, especially if attrition at follow-up is taken into consideration. Non-response is likely to be a particular issue for the control group.xxxviii xxxviii A self-completion survey as part of the Finnish basic income scheme achieved 31% response rates in the intervention group and 20% in the controls. The Mincome study in Manitoba achieved 54% of baseline interviews but with attrition of 20% and 12% in years 1 and 2 respectively suggesting a final response rate in the region of 40%. In contrast, the Scottish Household Survey achieved 64% in 2017, the Scottish Crime and Justice Survey 62% in 2017/18 and the Scottish Health Survey between 52% and 60% depending on the area.

78 To allow for non-response in this study, we would need to identify sites with larger populations than the initial sample size would suggest. Anticipated response rates in the order of 50% would require sites twice as big as the required sample size. This would greatly increase the cost of the payments in the pilot study.

For some outcomes, administrative data might be the best source. If so, non-response, non- participation and the resultant problems of bias and reduced power would be less of an issue, although consent to access data may still be required (see Section 7). However, as discussed earlier, there are limitations on existing administrative data in relation to the way they measure the primary outcomes (poverty, child poverty and unemployment) of interest, the scope to disaggregate data by sociodemographic group, the limited coverage of the population in a saturation site and the range of outcomes measured – it is unlikely they would cover all the secondary outcomes of interest in a CBI pilot. Therefore, the bulk of the data on outcomes would need to come from a bespoke survey, for which non-response will need to be factored into the sample size.

Whatever the sample size we ultimately seek to achieve, there will be a degree of fluidity in the relevant population because of incomers to and leavers from the communities selected during the course of the pilot study. We recommend that people moving into an area during the pilot should receive the CBI because not to do so might compromise potential community effects. We also recommend that they should be invited to participate in the evaluation. Additional questions could be included in the evaluation to identify the influence, if any, that the CBI had on their decision to move.

Consideration should be given to ways of mitigating ‘honey pot’ effects, where CBI is the primary reason for people to move into a pilot area. For example, it might be useful to establish residency rules so that people coming into the areas might not be eligible for the CBI straight away. We also recommend that new-borns should be included in the pilot.

We also suggest that people leaving an area during the course of the pilot should continue to receive the CBI for at least some time after their move. This may help to avoid the risk that the fear of losing the CBI discourages people who might otherwise move, for example, to new employment. Leavers would also continue to be included in the evaluation. Additional questions could also be asked of this group to see whether and how the CBI influenced their decision-making.

In summary, potential non-response means that we will need to oversample to achieve the required sample size. We anticipate a non-response rate of around 50% which means we would need to double the sample sizes estimated above.

6.3.4 Number and types of communities to include in the study The calculations in Table 11 estimate the sample size required to achieve adequate statistical power. We also need to consider how the choice of sites would affect whether and how we could generalize results to other communities.

One of the secondary outcomes of interest is ‘community effects’, that is, social or economic outcomes over and above direct, individual-level impacts, which may occur because all of the people within a defined area receive a CBI. Community-level effects include, for example, increases in volunteering, creation of informal caring networks, or the creation of new social enterprises, businesses and clubs. In Section 5 we recommended a ‘saturation’ model because we hypothesised that we would require a high proportion of the population within a community to

79 get the CBI if these community-level impacts are to be given their maximum chance to develop. We therefore need to understand the size and characteristics of communities that would make community effects more likely to occur.

We also need to identify a mechanism for identifying such communities and selecting those for inclusion in the pilot study. Interest has also been expressed in evaluating whether the impact of a CBI is different in different types of communities, for example, urban deprived, urban ‘average’ level of deprivation, or rural. Designing a study that would enable us to do that has implications for both the types of communities to include and the sample size. These are discussed in the remainder of this section.

6.3.5 Defining communities to explore community effects There is little evidence on how big a community needs to be for community effects to arise. We ideally need to identify geographical communities which include as many work, family, friends, caring, volunteering, club, socialising and market relations as possible. We also want to avoid selecting intervention or comparison populations that are exceptional in this regard as this will either reduce the generalisability of the findings or reduce the similarity between the intervention and comparison populations.

To help identify areas for inclusion in the study where it would be reasonable to expect community effects to occur, we considered two ways of classifying communities according to their size and characteristics. The first was the Carnegie UK Trust and the Scottish Towns Partnership Understanding Scottish Places (USP) methodology.xxxix This defines localities in terms of their independence from other areas, using information on the number of certain assets in the town, the diversity of the business and employment base in the town and the distance people resident in the town travel to work and to study. It is plausible that CBI is more likely to generate community effects in the more dependent and interdependent towns. Towns classified as such vary greatly in size from little more than 1,000 to the largest towns and cities in Scotland. This would suggest that even quite small communities might expect to see community effects. However, the evidence for this is limited. In addition, the USP system does not identify potential communities that could be selected from within cities.

The second was intermediate zones (abbreviated to ‘interzones’). There are 1,235 interzones across Scotland. Each interzone comprises a population of between 2,500 and 6,000. The boundaries reflect, as far as possible, naturally occurring neighbourhoods, whether by geography, transport links, housing types, etc. although the extent to which they represent a social community is still uncertain, as it is for the communities defined by the USP methodology.

However, interzones are a standard and accepted convention for reporting aggregated data, and data are likely to be more readily available at this level than for any other geography of similar size. We therefore recommend interzones as the basis for identifying intervention sites within and/or across a number of local authority areas. Sites could be contiguous interzones or in different parts of a local authority or located across a number of local authorities. Using interzones would suggest a minimum community size of 2,500 for inclusion in the pilot study (before taking into account non- response and attrition).xl xxxix https://www.usp.scot/StaticPage/Methodology xl A pragmatic way of keeping the cost of the pilot down would be to limit the selection list by interzone size. This might limit generalisability if there is a systematic difference between large and small interzones, but the trade-off would be worth considering if costs are considered to be a limiting factor.

80 We also recommend further work should be carried out to understand better the nature of community effects and the mechanisms by which they occur, both to help define the outcomes that would be measured in any pilot study, and to help define the size of community required. This work should draw on the insights and further work recommended in the Scottish Government’s recent report examining social connections within Scottish communities.xli

6.3.6 Types of community to include as study site(s) In addition to the size of area needed to generate potential community effects, we also need to decide on the characteristics of the intervention communities for the results to generalise to other areas in Scotland. For example, do we want the communities involved in the study to be similar to the Scottish average in terms of the primary outcomes (poverty, child poverty or unemployment) and/or are we interested in assessing the impact of CBI in more deprived communities where currently the primary outcomes are likely to be worse than the Scottish average? Likewise, we may want to explore the impact of CBI in both urban and rural communities.

Understanding how CBI would impact differently across different kinds of area is clearly of interest. Furthermore, spreading the intervention and comparison areas could help to mitigate against unintended contextual change that might impact on particular areas. For example, if an area were very dependent upon a single large employer, or if a new large employer entered an area, this could impact markedly on the local labour market and people’s incomes, confounding the potential impact of the CBI. However, the more types of community we try to include:

• for any given overall sample size, the smaller each community would need to be; • the less the power we would have to detect differences in outcomes between the communities of different types - if we want to draw comparative conclusions about each type of community, we need sufficient numbers in each; • the smaller the communities included, the greater the concern about whether community effects would be likely to occur; • clustering may affect the sample size required.

The sample size calculations above have not considered the impact of clustering on the required sample size. Peer reviewers of the interim report suggested it should be considered because the impact of clustering on required sample size can be substantial. If the difference between the intervention and control group were measured without taking account of the clustering of the data, the power of the study could potentially be overestimated, depending on the extent of the clustering effect. This would make it more likely that ‘significant’ effects were detected when in fact the actual power of the study did not warrant such a conclusion (see box).

The power calculations earlier in the report suggested that large sample sizes would be required in the pilot study, even without formally taking account of clustering in the power calculations. Taking account of clustering would likely increase the required sample size. Because each participant has to receive the CBI, there is a high cost per participant in the pilot study, so we need to be careful about any aspects of the study design that might increase the sample size even further.

xli Scottish Government. Social Capital in Scotland: report. February 2020. https://www.gov.scot/ publications/social-capital-scotland-measuring-understanding-scotlands-social-connections/pages/8/ Accessed 24 February 2020.

81 Clustering and sample size

Clustering means there is potentially less variation in outcomes between individuals within clusters than between clusters. Patients within a GP practice, pupils within a particular school or class, or as in this case, individuals within a particular community, are likely to have outcomes more similar to each other than to study participants in other clusters in the sample. The impact of clustering is typically to increase the sample size required.

The extent to which the sample size needs to increase (called the ‘design effect’) depends on the number of participants in each cluster and the ‘intracluster correlation coefficient’ (ICC). The ICC is given by the ratio of the between-cluster variance to the sum of the between- and within-cluster variances:

2 2 2 ICC = Sb / (Sb + Sw ) Where:

2 Sb = between cluster variance

2 Sw = within cluster variance This formula means that substantial within-cluster variation reduces the ICC, other things being equal, and therefore the design effect.

In this study, it is feasible that communities would display substantial within-cluster variation because any given community type may contain a wide variety of people in terms of measures of socio-economic status. This would reduce the ICC and therefore the design effect (other things equal). However, even if the ICC is very small, the design effect may be magnified by a large cluster size: increasing the number of clusters enhances power more efficiently than increasing the number of people within a cluster. A sample with a small number of large clusters, other things equal, would have less power than the same sample clustered in more, smaller clusters.

The clustering in the design proposed in this report arises for two reasons. The first is the saturation approach. If instead, the CBI were given to a random sample drawn from the general population, very few people would be drawn from the same community, reducing the clustering effect. However, we have proposed the saturation approach to enable us to uphold the universal principle and search for community effects.

The second reason clustering might arise would be the inclusion of different types of community. In Figure 4, for example, the intervention and control groups each comprise 1,600 people, based on the required sample size estimated earlier to receive a high rate of CBI (before non-response and attrition are taken into account). If we split the required sample between, say, the three community types (urban average, urban deprived, rural average), we would have 533 people in each type. This is much less than the sample size needed to have enough power to look at the differential effects of the CBI within each community type. In addition, even if we limited our analysis to comparing the overall effect size across the three types of community as a whole, we would need to take account

82 of the clustering in the sample size calculation, and the design effect for doing this might be large due to the small number of large clusters.

This would not be as problematic for the high payment group, where we estimated the required sample size to be 1,600 (or 2,500 minimum if we base our community size on interzones in order to be able to detect community effects). This would mean that achieving this sample size in each of three community types might be feasible (unless the design effect is very large). But it would still be challenging for two reasons: first, 2,500 is a minimum community size and, second, 2,500 is the required sample size for the high payment group, where a large effect size is expected. In the low payment group, we estimate needing 14,600 to get adequate power. A sample this large in each of three types of community would be very expensive and for that reason, unlikely to be feasible.xlii

Having multiple communities of different types would therefore be difficult. Whether the rationale is to test CBI effects on different types of community, or to create a sample powered to allow for the clustering that arises from including multiple community types, the impacts on the required sample size and cost are likely to be prohibitive in the context of a CBI pilot.

Conversely, if we select two pilot communities (one each for the high and low payment) and a control group, we would not need to consider design effects arising from clustering, we would not need to inflate the sample size and the existing sample size calculations would apply. However, the size of the interzones is a practical constraint on this. As noted earlier, the population of the interzones ranges between 2,500 and 6,000. Therefore, in the high payment group, the community selected may have more than the 2,500 people required, which means the statistical power would be sufficient but would add to the cost of the pilot payments (and the evaluation). In the low payment group, the sample of 14,600 would need to be made up of a number of interzones, introducing the risk of clustering effects and also the risk that the sample size exceeds the number required, with the additional costs that brings. The design effects of the clustering on the required sample size and the additional costs would need to be explored if and when the sample for the study population were being drawn.

Because we are not expecting to see community effects in the control group, the control group could be constructed by drawing a random sample from the population in the communities remaining in the sampling frame after the pilot areas have been selected, stratified to ensure it is similar to the Scottish population in relation to the primary outcomes. This would reduce the potential for site-specific factors, such as closure or opening of a large employer, to impact on the outcomes of interest. It would also avoid any risk of spill over effects between adjacent intervention and control areas contaminating the results.

The outstanding methodological concern would be whether representative communities can be selected for the pilot sites that would provide results generalizable to the population as a whole.xliii A single site inevitably limits generalisability to a degree but this could be mitigated through the criteria drawn up to identify the intervention communities and control group selected to take part in the pilot. These are discussed further below.

xlii Note that these are the minimum sample sizes before taking non-response and attrition into account. xliii Stratification in a general population sample could be used to try and achieve this, but as noted above, the small number of people within individual communities would make it unlikely that there would be any community effects. 83 Figure 4: Clustering in a Pilot Looking at the Impact of a High Level of CBI Across Different Community Types

Intervention Group (receive the CBI) Control Group (status quo) e.g. n e.g. n = 1600 = 1600

Urban deprived Urban deprived e.g. n = 533 e.g. n = 533

Urban average Urban average e.g. n = 533 e.g. n = 533

Rural Rural e.g. n = 533 e.g. n = 533

Note: Assumes required sample size n=1,600 from the power calculations in Section 6.3.2, not taking into account the detection of community effects, non-response or sample attrition.

6.3.7 Criteria and process for identifying and selecting communities for inclusion in pilot study The discussion so far has highlighted the need to consider and agree the number and type of communities to be included in the pilot study. This section suggests the criteria and process for identifying and selecting the communities to be included in the sampling frame from which the pilot areas and control group would be randomly chosen.

Work undertaken by Fife council during the feasibility study developed a process for creating a sampling frame by identifying interzones within predefined percentile bands using data from the Local Level Household Income Model Estimates. The process identified interzones either within a 47.5-52.5 percentile band or a 45-55 percentile band around the Scottish median based on the proportion of the population in household poverty. The method was also used to identify interzones within +/- 2.5 percentile band and a +/- 5 percentile band around the 80th percentile, demonstrating how the method can be used to identify areas facing greater economic hardship.

We suggest applying this approach using the most up to date data available at the point any CBI pilot study is undertaken. We also suggest that selection should be based upon data on the three primary outcomes, or indicators related closely to them, given their importance to the study. The percentile bands could be set to identify either interzones close to the Scottish average or close to a predefined threshold towards the high end of the poverty/unemployment distribution if stakeholders wish to conduct the pilot in more deprived areas.

If and when it was agreed that a pilot study should go ahead, potential data sources would need to be explored to ensure the most up to date information was used. For the thresholds chosen, the interzones falling within these thresholds for the three primary outcomes would need to be identified. A decision would also need to be taken as to whether the interzones included in the sampling frame needed to fall within these thresholds for each of the three outcomes. In practice they are likely to be related, i.e. interzones with high levels of unemployment may well have high

84 levels of poverty and child poverty, which may mean that the decision about whether interzones need to meet one, two or all three criteria do not make much difference to the number or range of interzones included in the sampling frame.

It is unlikely that this process will identify areas that perfectly match Scotland as a whole, either in terms of the levels of the primary outcomes or the degree of variation across the interzone areas. In addition, the areas identified will differ in relation to criteria not used to define the sampling frame but which may affect the outcomes of interest in the wider evaluation. This might include substantial new housing, infrastructure or industry development, which could mean that the effects of CBI are confused with the other factors. As a result, criteria to identify and exclude areas with substantial other developments that would interfere with the evaluation would need to be developed and applied. However, the criteria defined above will ensure that outliers in relation to the three primary outcomes will not be included in the sampling frame. Also, generalisability would be enhanced through the process evaluation and through analysis of outcomes across different subgroups within the areas chosen.xliv This would provide information on the extent to which the impacts of the CBI might differ between groups and the possible reasons why.

6.4 Process Evaluation

It was noted earlier that process evaluation helps us to understand not only whether an intervention is effective, but also for whom, how and under what circumstances. This includes understanding how the local social and economic context shapes the impact that an intervention might have. We have, therefore, suggested that the evaluation of the pilot CBI, if progressed, should include a process evaluation.

Adapting MRC guidance,50 we would suggest that a process evaluation should cover four main areas: implementation, beneficiaries’ experience of service delivery, mechanisms of change, and wider system/context. These are discussed briefly below, together with suggested methods for data collection.

6.4.1 Implementation of a CBI This would involve learning from setting up and delivering the CBI, including integrating the CBI within existing benefit delivery systems, and local support services. This could be undertaken through semi-structured interviews with the relevant delivery agents and local partners, e.g. JobCentre Plus, welfare/debt advice, relevant DWP and HMRC officers and policymakers. Repeat interviews would also be useful to explore whether the CBI was implemented as intended or whether it was modified, adapted or changed in the course of implementation. This could also explore the reasons and implications of these adaptations, for service delivery, recipients of a CBI and the wider system.

A process evaluation should also explore reach, i.e. who actually receives the CBI. In the context of a pilot CBI based on a saturation model this would be primarily about monitoring the uptake of the CBI by all those ‘normally resident’, including those who move into a site in the course of a pilot. This could primarily be through routine data collection by the CBI delivery service.

A key issue raised during the evaluability assessment, by peer reviewers, DWP and other xliv Bonell C et al, Assessment of generalisability in trials of health interventions: suggested framework and systematic review. BMJ. 2006 Aug 12; 333(7563): 346–349. doi: 10.1136/bmj.333.7563.346. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1539056/ 85 stakeholders, is the management of transitions on and off the CBI, in particular at the end of the pilot. Mechanisms will need to be in place to support people at both the start and end of the pilot period (if recipients are no longer given the CBI is withdrawn at that point). In addition, the pilot will need to be responsive to changes in individual and household circumstances over the pilot period. If delivered on a ‘saturation’ basis, then transition support will need to be extended to people coming into the pilot sites as well as those moving out with the pilot areas. The recommended mechanisms are set out in section 7.2. Although these mechanisms are an artefact of the pilot nature of the intervention, a process evaluation should explore how they operate in practice since they may impact on observed outcomes. This might occur if, for example, concerns about transitioning off the CBI at the end of the pilot affects people’s decisions to take up work or training or to set up new enterprises.xlv

6.4.2 Experience of delivery To assess the secondary outcome, ‘improved experience of, and relationships with social security and welfare support services’, the process evaluation could explore beneficiaries’ experience of, and views on, the delivery of the CBI, including, for example, the frequency of receipt, the means of delivery, who within a household receives/manages the payment (including who manages money for children and people who lack capacity), the amount received, the implications for access to and take up of other entitlements (including ‘passported’ benefits), and other implications such as whether and how the level and payment delivery options of a CBI help or hinder families’ ability to manage debt

It would also be important to explore the implications of removing conditionality, contrasting it with the views of those in the control group who continue to face conditions for the benefits replaced by the CBI.

6.4.3 Mechanisms of change A process evaluation should explore whether and how receiving a CBI affects some of the circumstances that the theory of change suggests might be influenced by a CBI and which, in turn, might affect longer term social and economic outcomes, for example:

• Financial circumstances and financial decision making • Relationships • Life events and life choices (including work/labour market participation, education, health, creativity, caring etc.) • Community and neighbourhood activity.

Qualitative evidence on both experience of delivery and the mechanisms of change could be obtained through in-depth interviewing – potentially via purposively sampled panels who would be followed up over the course of the study period. The use of financial diaries could be considered, however, these are very intensive in terms of demands on individuals and in interviewer/researcher time.51, 52

6.4.4 Wider system/context Evidence on external factors that may influence the implementation and mechanisms of change could be collected through the delivery agents and relevant partners, but also other relevant local xlv The governance implications are discussed in Section 7. 86 agencies, such as Chambers of Commerce, Economic Development Partners, or local poverty commissions (where applicable).

Interviews or focus groups with such agencies could explore the implications of a CBI for wider community infrastructure such as education, training, social work, health, third sector, community groups, which in turn could help to develop understanding of the extent and nature of community effects. This would exploit the saturation site methodology to explore the implications for a community as a whole, over and above the direct effects on the individuals who live in the area. A variety of methods could be used, such as participant observation or other ethnographic methods, bearing in mind that many such methods are relatively resource intensive both in terms of gathering and analysing the data, relative to the numbers of participants involved.

This raises the question of evaluation costs, considered in the next section.

6.5 Evaluation costs

The evaluation proposed would require substantial resource over a number of years. In considering the potential costs, it is important to bear two things in mind. First, the evaluation costs will be a small proportion of the total costs of the pilot given the high cost per person of involvement in the pilot itself – each participant will receive the payment at either the high or low level so the direct costs of the pilot will increase in direct proportion to the sample size. Second, the costs of an outcome and process evaluation of the kind recommended in this report relate to variables that we are not yet in a position to finalise. Therefore, we can only give a broad indication of the potential costs.

The main drivers of the evaluation cost include:

• The size of the intervention and control groups (including the sample boosts to account for non- response and attrition) • The means of data collection - for example, self-completion online surveys are relatively low cost but limit the types and richness of the data that can be collected; computer assisted personal interviewing conducted face to face is relatively expensive but has a number of advantages, including faster interviews, high quality data in part due the scope for automatic edit checks and a quick flow of information from the interviewer to the survey database. Collecting and analysing qualitative data would be more time/resource intensive than survey data in terms of data cost per person but would typically be carried out on much smaller samples and would allow themes generated in the data collection process, including themes generated by respondents themselves, to be explored in more detail. • Set up costs which depend in part on the different data collection tools to be designed • Frequency of data collection. As a minimum, it would be important to collect data at baseline and follow up. Data collection at intermediate points would also be of interest but the cost and logistical implications would be substantial. • The amount of data to be collected: longer surveys take longer to administer and require more time for data checking and cleaning. Survey length depends on the number of outcomes on which data are sought (see Section 6.2).

Because decisions still need to be taken about these cost drivers it is only possible to give indicative costs at this stage. For example, the approximate overall cost for the current Scottish

87 Crime and Justice (SCJS) survey is £5.3m for six waves of data collection over six years.xlvi The study aims to achieve a sample of 6,000 adults each year, using an unclustered but stratified sample, and aims for a response rate of 68%. That has not quite been achieved in recent years – response rates in 2017/18 were 62.4% generating nearly 5,500 completed interviews.xlvii If these response rates were maintained over the duration of the study, the cost per completed interview would be around £160 per respondent. The study uses Computer Aided Personal Interviewing (CAPI) and Computer Aided Self Interviewing (CASI), and interviews last on average around 40 minutes.

The Scottish Household Survey achieves a sample of around 10,500 respondents a year. The survey uses computer assisted personal interviewing and takes place face-to-face in people’s homes. Each interview takes around 40-60 minutes. The survey costs around £2.5 million a year or around £240 per respondent.

These examples illustrate the scope for variation in methods and costs. The average costs in these surveys would obviously not translate directly into the costs if a survey were carried out as part of a CBI evaluation – some will represent fixed costs that don’t increase in direct proportion to the size of the survey. In addition, the same methods would not necessarily be used, and interview length might differ, depending on the number of outcomes. However, they indicate the likely order of magnitude of the cost of a survey of this scale.

A survey of 17,100 respondents plus a similar number of controls would require in the region of 34,000 interviews at an indicative cost of £200 per respondent suggesting a cost of £6.8m for each wave of a survey, or £13.6m for a study with a baseline and follow up. The costs of the process evaluation would need to be added to this, which we have not estimated due to the wide range of approaches that could be adopted. Staff time to undertake analyses of routine data would also need to be protected, perhaps from amongst existing public sector evaluation teams. Reference was made earlier to the potential value of collecting data at an intermediate point in the pilot study, but these figures give an indication of the high cost of additional waves of data collection.

The costs of the evaluation could be reduced in a number of ways, for example, by not powering the study to detect differences in impacts between men and women, by compromising on the power of the study in general, by reducing the number of outcomes measured in a survey or by limiting the evaluation to a process evaluation based on qualitative methods rather than a survey. But each of these reduces the value of the evaluation in some way so a decision needs to be taken on whether the additional costs of a more extensive evaluation are justified by the additional benefits.

6.6 Conclusions and recommendations

6.6.1 Conclusions Overall, we recommend that a randomised controlled outcome study with two intervention arms (one receiving the high payment, the other receiving the low payment) and a control group offers the best way of understanding the potential impact of a CBI on a range of social and economic outcomes. However, there are several trade-offs that need to be made in choosing the best approach to the evaluation. xlvi Public Contracts Scotland Award Notice: https://www.publiccontractsscotland.gov.uk/search/show/ search_view.aspx?ID=SEP221786 xlvii Scottish Government. Scottish Crime and Justice Survey 2017-2018: main findings. 26 Mar 2019. Available at: https://www.gov.scot/publications/scottish-crime-justice-survey-2017-18-main-findings/ pages/5/ Accessed: 18 Feb 2020.

88 There is a tension between having enough statistical power to assess change in the primary outcomes, the desire to have multiple pilot sites in different types of community with each large enough to generate community effects, and the potential cost of the pilot and evaluation. For any given sample size, the more community types we seek to include, the less the power to detect differences in outcome by community type, and the greater the concerns about the potential to detect community level effects.

Increasing the overall sample size would help to address this problem but the greater the size of a pilot, the higher the direct CBI payment, administrative and evaluation costs would be. Keeping down the costs of the overall pilot requires either the community size or the range of community types to be reduced.

In summary, we conclude that the most appropriate approach to evaluation would be:

• an outcome evaluation • in the form of a randomised controlled trial • randomising communities to each of the high payment and low payment CBI options drawn from a sampling frame of eligible communities based on defined criteria • delivered on a saturation basis • including a control population comprising a stratified random sample of the population drawn from the same sampling frame as the pilot communities • complemented by a process evaluation exploring the mechanisms by which any change in outcomes came about, including any facilitators and barriers to the successful implementation of the CBI and any differential effects between different groups of recipients.

We acknowledge the limitations of the proposed pilot and evaluation, in particular:

• the limits on generalisability to other areas • the limited scope to explore the role of contextual effects due to the limited range of contexts in which the CBI would be implemented • the risk that single sites might be contaminated by economic shocks peculiar to those areas.

However, there are benefits to these proposals that mitigate these risks, in particular:

• the lower costs compared to a study in multiple communities of different types where clustering and reduced sample size in each type would reduce the power of the study • the greater potential to explore community effects compared to a pilot using a random sample of the general population • the adherence to the principles of a universal CBI.

Perhaps the most important factors to consider in deciding whether to go ahead with the pilot and evaluation proposed are not specific to whether the chosen approach is based on a random sample of the general population or saturation approaches based on one or more communities. Rather, they relate to the wider questions raised at the start of this section about whether any feasible pilot will provide a valid test of the potential benefits of a CBI given:

89 • that some of the potential impacts of a CBI are unlikely to occur because of the time-limited, geographically-narrow focus of the proposed pilot • the institutional barriers to the pilot proposed in this report, discussed in Section 8.2, that could not be overcome in the foreseeable future.

These are important limitations of a pilot accompanied by an outcome evaluation on the scale and using the design proposed in this report, but they need to be considered against the strengths and weaknesses of the alternatives – in particular a smaller scale pilot and/or a modelling approach. These arguments are summarised in Table 12 below.

90 Table 12: Strengths and Weaknesses of Different Approaches to the Evaluation of the CBI

Study design Strengths Limitations Micro pilots (i.e. pilots • Low cost of pilot and evaluation • Unlikely to be an evaluation of a CBI that adheres to delivered in a small • Feasibility (although the type of CBI that could be the principles set out in the report geographical area or to piloted and whether participants could receive • Unlikely to induce the behavioural effects in enough particular sociodemographic the CBI as defined in Section 5 would depend on people to generate the economic impacts of a CBI group) whether and how the institutional barriers discussed in practice in Section 8.2 could be resolved. It may not be • Not possible to measure outcomes quantitatively feasible to change benefit or tax rules for a small due to small numbers number of people.) • Not possible to measure community effects Modelling • Low cost of evaluation - no pilot involved • Weak evidence exists on which to base model • Possible to model much wider range of scenarios parameters. and longer-term outcomes at low additional cost • Modelled behaviour and impacts may not mimic actual behaviours and impacts due to unanticipated or imperfectly understood effects. • No scope to explore implementation or other process variables • This would add little to what is already known through existing modelling of CBI by various groups Random sample, general • Wide range of community types can be represented • High cost population in a sample, although not with sufficient power to • For given sample size, few people in each draw type-specific conclusions regarding impacts geographical area – less scope to explore unless sample sizes are very high. community and contextual effects • Results unlikely to be contaminated by community- • Not universal. specific economic ‘shocks’ • Provides some (albeit imperfect) empirical evidence on individual responses to CBI to guide policy and to inform modelling of longer-term impacts

91 Study design Strengths Limitations Randomly chosen multiple • Adheres to the principles of a universal CBI • Highest cost if factoring in design effects due to saturation sites • Potential to explore community effects compared clustering and/or seeking to power study to look at to a pilot using a random sample of the general community type-specific impacts population • Provides some (albeit imperfect) empirical evidence on outcomes to guide policy and to inform modelling of longer-term impacts • Wide range of community types can be represented in sample, subject to cost constraints (see limitations) – would enable exploration of contextual influences on outcomes. • Less prone (than single-site option) to economic shocks ‘contaminating’ the results Randomly chosen single • Adheres to the principles of a universal CBI • High cost saturation site • Potential to explore community effects compared • Limited scope to explore the role of contextual to a pilot using a random sample of the general influences on outcomes due to the limited range of population contexts in which the CBI will be implemented. • Provides some (albeit imperfect) empirical evidence • Risk that single sites might be ‘contaminated’ by on outcomes to guide policy and to inform economic shocks peculiar to those areas modelling of longer term impacts

92 6.6.2 Recommendations a) The theory of change should be developed over time, and revisited as new evidence becomes available, new stakeholders are identified and/or as stakeholders identify new areas of interest. b) Based on the theory of change, the primary outcomes should be changes in poverty, child poverty and unemployment. Secondary outcomes should include community level social and economic effects, improved health and well-being and improved experience of the social security system, with the final list developed over time as the theory of change evolves. c) The most appropriate and robust study design would be a (clustered) Randomised Control Trial (RCT). The intervention groups should comprise one or a small number of communities selected at random from a sampling frame comprising a list of communities drawn up on the basis of agreed inclusion criteria. The control group should be a stratified random sample of the population drawn from the same sampling frame as the pilot communities. d) Everyone in the intervention community or communities should get the CBI: in line with the ‘saturation approach’ suggested in Section 5. e) Within the pilot community/communities, everyone receiving the CBI would be invited to take part in the evaluation. Drawing a sample from a group receiving the CBI would greatly increase the size of the intervention group required and make the pilot prohibitively expensive. f) One community should be selected to receive the high payment and one (or more depending on the size of the interzones) to receive the low payment, together with a control group. Sampling from a range of types of community would increase the number of people required in the pilot study and substantially increase its cost. g) The communities selected to receive the CBI should be based on ‘interzones’. h) A minimum sample size of around 800 at the high rate of CBI and 7,300 at the low level of CBI would be required for the evaluation to be adequately powered. To be able to detect different effects for males and females separately would require the minimum sample sizes to be doubled to 1,600 for the high level of CBI (or 2,500 minimum if we base our community size on ‘interzones’ and we want to be able to detect community effects) and 14,600 for the low level. i) Oversampling should be carried out to address potential non-response and attrition, although this will increase the costs of the pilot in terms of CBI payments. j) People leaving and people entering the study area should be included in the pilot, with consideration given to various eligibility criteria to reduce the risk that differences in the availability of the CBI between study and neighbouring areas affect the outcomes by distorting decisions regarding uptake of employment, moving area etc. Children born in the study area during the study period should also be included in the pilot and receive the CBI. k) A process evaluation should be carried out alongside the outcome evaluation. This would provide information on how the CBI is implemented, who it reaches and the barriers and facilitators to achieving outcomes. l) A number of differentdata sources should be used to collect data on outcomes and processes (subject to necessary consents and data protection issues). These include administrative

93 data routinely collected as part of administering current taxes and benefits and as part of administering a CBI; ‘bespoke’ survey data and qualitative data designed for the purposes of the CBI evaluation. m) Further work should be carried out to understand better the nature of community effects and the mechanisms by which they occur, both to help define the outcomes to be measured if is agreed that a pilot study should go ahead, and to help define the size of community required. n) The recommendations above specify what we believe to be the most robust evaluation possible. However, they also highlight substantial challenges and important limitations of a pilot and outcome evaluation on the scale and using the design proposed in this report. These need to be considered against the cost of the pilot and evaluation and the strengths and weaknesses of the alternatives – in particular a smaller scale pilot and/or a modelling approach.

94 Section 7: Policy Pilot Governance and Research Ethics

7.1 Introduction

Policy pilots have been described as ‘test runs’, the results of which will help to influence the shape and delivery of the final policy. They can be used to test the possible effectiveness and/or the practicalities of implementing a policy prior to full implementation.53 As such, process and/or outcome evaluation (and monitoring) are integral to policy pilots and the design of the pilot and the design of the evaluation are inextricably linked.

As discussed in Section 4, one option for a Government may be to roll out a policy, such as a CBI, without trialling it in advance, perhaps because a pilot would have no chance of influencing policy or delivery in the time available, or because it is based on prior manifesto commitments.53 Contingent on the nature of the intervention, this may raise its own ethical questions if a whole population were subjected to an untested intervention. It may also waste resources if it turns out to be ineffective and is abandoned. The alternative would be to trial the intervention prior to full implementation. By its very nature this implies some sort of selection and comparison because only some areas, populations groups or individuals, would be included in a pilot. Others would be excluded. For the purposes of evaluation, having an intervention group and a control group allows for comparison of outcomes.

There are a number of different evaluation designs for constructing intervention and control groups. In ‘natural experiments’, for example, inclusion or exclusion may be the result of a naturally occurring difference in exposure to an intervention not under the control of a researcher.54 A ‘quasi- experimental design’ involves an ‘administrator’ such as an official or policy maker allocating people to an intervention or comparator group.55 In a randomised controlled trial, a researcher allocates people to one or either group at random. As discussed in Section 6 above, for the purposes of an evaluation of a pilot CBI, it is recommended that the CBI is delivered to all the members of a community (a saturation site) selected at random. Random allocation is recommended because it is the most robust method of preventing selection bias due to systematic differences between the intervention and control groups which are likely to affect outcomes.56

Inclusion, exclusion or differential exposure to a new policy or intervention, does, though, raise questions relating to both policy governance and research ethics. These are particularly pertinent to a CBI pilot where, in relative and absolute terms, treatment and ‘control’ groups may be differentially affected financially, or in kind (e.g. through exemption from meeting certain conditions in order to receive welfare benefits, or through exposure to greater or lesser risk of direct or indirect detriment or harms). Any pilot that was implemented or evaluated in ways that were unethical would be infeasible.

To explore these issues the following sets out:

• The implications of implementing and evaluating a policy pilot based on giving some people/ communities a CBI (the intervention group) and comparing the outcomes against a control group who do not receive a CBI. This is examined through the concept of ‘equipoise’.

95 • The implications for pilot governance, in particular, the need to protect the intervention group(s) from harms they would not be exposed to if they were not part of a pilot. This includes assessing and mitigating potential negative differential impacts based on a protected characteristic or socio-demographic status. • The implications for research ethics and governance.

7.2 Implications of a control trial: protecting intervention and control groups from relative harms

7.2.1 Equipoise As noted above, if the pilot proposed were to progress, the aim would be to evaluate the impact of providing individuals with a regular income (at a high or low level) which they receive without having to meet conditions or eligibility criteria (other than residence within the intervention area). In order to do this the suggested evaluation design is for a controlled study with an intervention group receiving the CBI and a comparator group.

In clinical research the principle of ‘equipoise’ is applied where there is genuine uncertainty whether a treatment is beneficial. It means that no participant in a controlled trial is knowingly given inferior treatment. The principle follows from the duty of beneficence (a healthcare professional’s commitment to care for a patient as best they can). As a concept, it has been suggested that it may not apply directly to social, public health or preventative interventions for populations of ‘healthy’ subjectsxlviii – though a principle of ‘fairness’ may apply if a public health need is unmet.57 Further, for social interventions there may not be the evidence around outcomes, or there may be multiple outcomes around which there is uncertainty in relation to some but not others.58

The question for this feasibility study is whether, based on current knowledge, we have any reason for thinking that if a pilot CBI were to be trialled, the control group would necessarily be worse off than the group who received the CBI (and were exempted from conditions related to certain social security benefits)? Under normal circumstances it would be assumed that additional money would be beneficial, particularly for people on low incomes. This would mean, in the case of the CBI (assuming recipients experience no-detriment), the control group are, by definition, likely to be at a financial disadvantage.

Conversely, given the limited comparable evidence from other studies, there may be sufficient uncertainty in relation to the wide range of intended and unintended outcomes (positive and negative), to believe that the control group (and in fact the wider population) is not being denied a ‘treatment’ known to be of benefit. Even if the benefit of a CBI in direct financial terms is clear, the uncertainty in relation to the wider impacts for the individual and society, may justify the argument that the intervention and control groups in a trial of a CBI would be in equipoise. Arguably this extends to the proposal to pilot two different payment levels; insofar as there is sufficient uncertainty as to the wider impacts of a CBI, piloting varying rates may not undermine the argument for equipoise.

There is a tension here between needing to protect (individual) participants (intervention and control groups) from absolute and comparative harm and generating robust evidence to inform future decision making, including for rolling out an intervention in a way that maximises benefits and xlviii The author argues that ‘A trial in which the preventive effects of two interventions are compared can be morally acceptable, even if there is no equipoise regarding these arms of the study.’ (p.151). 57

96 minimises harms to a wider population. Given the degree of uncertainty relating to a number of the individual, community and social outcomes identified as of interest, it could be argued that:

• There is genuine uncertainty about the potential effects of a CBI on poverty and income in the context of the current social security system (for those who receive welfare benefits); • There is less uncertainty that improving incomes and reducing poverty is likely to improve health and wellbeing outcomes • There is uncertainty about individual non-health outcomes • There is uncertainty about community level outcomes

Uncertainty regarding the wider macro-economic outcomes will not be resolved by a pilot alone. As noted in Section 6, this will make learning from any pilot scenario less complete than we would like.

Given the variable degrees of uncertainty across a number of individual, community and social outcomes our provisional suggestion is that, based on the design of pilot proposed, an RCT would be ethical on the grounds of equipoise. It is however recommended that, if a pilot were to proceed and be evaluated, a submission is made to an appropriate research ethics body (see section 7.4.2 below). This would need to be based on the model that it is proposed to trial and the proposed evaluation design. The Scottish Government may also wish to explore with potential participants how they might view the relative benefits and harms, and the trade-offs between individual and societal outcomes. This could be part of a consultation process prior to implementation of a pilot.

Equipoise arises in a context of selection; it implies that something is given to one group of individuals and deliberately withheld from another in order to assess whether any impacts are due to the intervention or would have occurred anyway (‘the counterfactual’). As noted above, for the purposes of research, the ethical basis for allocation and exclusion is that there is genuine uncertainty about whether the new intervention actually produces better outcomes than ‘usual practice’. If it progresses, the selection and exclusion of some populations, groups or individuals for the purposes of trialling the CBI policy would also need to be administratively just. That is, the process needs to be within the legal competence of policy makers and the processes of allocation or exclusion must be seen to be reasonable and procedurally fair. It may also be necessary to demonstrate that it is not incompatible with EU Law (where that applies) and is not in breach of rights under the European Convention on Human Rights.

A process of randomisation of areas (if the CBI were delivered on the basis of saturation sites), based on explicit, defensible and fairly applied criteria may support arguments that selection is procedurally fair and reasonable. We recommend, however, that the Scottish Government seek expert advice on the legal and procedural basis for a pilot if and when a decision is taken to go ahead with a pilot. This again would need to be based on the model that it is proposed to trial. Parliamentary scrutiny processes (UK and, or Scottish) would obviously apply where legislative changes are required.

7.2.2 Protecting the intervention group from absolute harms Equipoise and administrative justice are aimed at ensuring that ‘excluded’ groups are not at detriment compared with intervention groups. That is they are not unfairly, or arbitrarily, excluded from access to an intervention that would be of greater benefit to them than their ‘usual treatment’ (e.g. receipt of existing welfare benefit entitlements or application of current tax rates). Research and policy ethics also require that the intervention groups are protected from absolute harms which

97 they would not otherwise be exposed to if they were not part of a pilot. This is particularly important if the model of CBI piloted, as proposed in Section 5, means that some welfare benefits will be replaced by a CBI, but within the context of the current social security system. Welfare benefit recipients need to be protected from any unintentional harms that might result from the (temporary) replacement of some existing benefits. Consideration also needs to be given to avoiding future detriment if, for example, the pilot is not subsequently rolled out, or people leave an intervention site before the end of the pilot. It would also be important not to expose groups already vulnerable to disadvantage due to stigma, prejudice or discrimination, to greater comparative risk. The following outlines the mitigations recommended to minimise these three sources of absolute harms.

7.2.2.1 Principle of no detriment

A fundamental principle informing the feasibility study is that participants in the intervention sites should not experience (financial) detriment compared to individuals not involved in the study. The principle of no detriment could also be expanded to include institutional detriment associated with a reduced or lack of access to support services due to participation in a CBI pilot. Such services may include loss of access to Jobcentre Plus if some existing benefits which provide access are suspended during a pilot study.

The principle of no detriment is important for two main reasons. Ethically it would not be appropriate to recruit participants (whether participation was compulsory or voluntary) who would be in detriment compared to people not involved in the pilot. Secondly, to consider compulsory participation as an ethically feasible option, it will be necessary to ensure there is no harm or detriment to participants.

In relation to interaction with the social security system, commissioned research by CPAG explored in detail the implications of receiving a CBI for participants who are in receipt of means-tested and non-means tested benefits. This includes consideration of whether and how receipt of a CBI may impact on other benefits over the period of the pilot as well as in the longer term.

Within our preferred model there are a number of benefits proposed to be suspended during a pilot. The low level of CBI proposed is at least the level of the benefits replaced. However, some of the suspended benefits allow passporting to other entitlements and elements for financial and service support. To prevent detriment to participants, we suggest that these additional elements are retained for participants during a pilot. These include passported benefits, Jobcentre Plus support and National Insurance Credits (which if lost could mean participants might lose future entitlement to benefits for which national insurance contributions are required).

There is also a risk that participants receiving legacy benefits at the start of a pilot face being transferred to Universal Credit at the end of a pilot. This would likely cause participants to lose out on transitional payment protection which they may otherwise have received if they were not involved in a pilot. Participants would therefore be at risk of relative and actual detriment compared to those not involved in a pilot. Some participants may be unwilling to participate if they are automatically moved to Universal Credit upon finishing a pilot as some may be worse off financially.

We have proposed that the low rate CBI for those of pension age is set at the same level as the new state pension (£168.60). The Steering Group recognise the likely complexity of CBI interaction with the variations of state pension entitlements and that pension age participants will have a range of existing payments, premiums and top-ups. To help avoid detriment, it will be important to ensure participants receiving payments in excess of the new state pension rate continue to have access to

98 this amount and those on lower incomes continue to get any passported entitlements.

It is also likely that suspension of State Pension during a pilot would involve participants being asked to defer entitlement to their pension. State Pension can only be deferred once during a lifetime. Participants would therefore lose the opportunity to defer their pension entitlement at a later date.

The risk of financial detrimental is closely linked to the pilot model design. For example, the risk of losing out in absolute terms during a pilot depends in part on the level of CBI and whether CBI is regarded as income for the purposes of assessing eligibility for means-tested benefits. If CBI was disregarded as income for the calculation of means-tested benefits this would reduce the risk of detriment through the loss of benefit entitlement for participants and through the loss of passported benefits.

The risk of financial detriment is of most concern to low income participants or those in receipt of benefits. However, there may be circumstances where the design of a CBI pilot does lead to financial detriment, particularly one that seeks to mirror the national fiscal policy required to finance a nationwide CBI policy. It is likely that to fund a CBI of the form proposed in the pilot, the income tax system would need to be substantially more progressive, taxing higher earners at a higher rate than the current situation. Doing this as part of a pilot could be ethically problematic and raises questions about whether a government can mandate individuals to participate. Therefore, although not currently being proposed by the Steering Group, careful consideration would need to be given to any experiment incorporating tax changes of this kind.

We recommend that participants, particularly those who are vulnerable and/or on low incomes, should not experience detriment (financial or otherwise) compared to individuals not involved in the study. The risk of detriment can be reduced by ensuring CBI payments are disregarded as income for means-tested benefit calculations. Preventing detriment will require careful design of the CBI, the full cooperation of DWP and HMRC, and may involve mitigation in the form of compensation or additional support.

7.2.2.2 Managing transitions

A pilot, if it goes ahead, is anticipated to be time-limited, with implications for people once the pilot ends (assuming it is), and payment ceases. The issues that need to be addressed within a discussion of transitions are:

• The principle of ‘No detriment’ needs to include management of the risks pre-pilot, during the pilot and post-pilot. Examples of detriment include loss of net income due to changes to tax and benefits during the pilot stage and detriment arising from changes to accessing services. • Processes to deal with individuals’ circumstances that change over the course of a pilot and would need to be reflected both in support to obtain relevant benefits (and/or to manage any potential tax implications) during the pilot period and in transitions back to the wider social security (and tax) system (if a pilot were not rolled out). • The support provided to participants before, during and after a pilot. For example, the support available at the end of pilots to access benefits or manage the loss of extra income that CBI may have represented. • Knowledge and awareness of the pilot at the start and throughout the study.

99 • Behaviour changes brought about, for example, through pilots changing saving and spending patterns such as reducing spend towards the end of a pilot or increasing savings or increasing credit at the start of a pilot on the grounds of a guaranteed basic income for a medium term.

There is a need to ensure that appropriate time is allocated to planning the implementation of a local pilot and a transition strategy to support participants both on to and out of the basic income pilot.

For all participants, consideration will have to be given as to how to make them aware of the likely time-limited nature of the intervention, and to offer access to welfare benefits advice and financial management services, including where appropriate, the management of credit arrangements or debt. Consideration could also be given to a tapered reduction in CBI following the end of a pilot. This would need to be planned in advance as part of the preparations for any local pilots of basic income in Scotland.

The proposed framework for transitions and support is illustrated in Figure 5.

Figure 5: Transitions and Support Framework

Pre-pilot Transition Pilot Transition Post-pilot Assessment Recruitment Evaluation Support On Phase Off Support Change of Circumstance

The phases outlined above are explained more fully in Table 13 below. In each of these stages, participants are likely to require differing degrees of support, but this carries with it risks in terms of how it might influence the impact of the CBI and therefore the findings from the evaluation. Any special arrangements put in place to avoid detriment or manage risks of need to minimise the risk of ‘contaminating’ the evaluation by influencing outcomes in a way that would not occur were the CBI rolled out nationwide.

Table 13: Levels of Support

Level Support Universal Support available to all Additional Additional support based on individual needs Intensive Intensive interventions based on individual needs

Table 14 explores each stage in more detail and provides examples of the activities expected to be undertaken at each stage.

100 Table 14: Transition Support Examples

Stage Definition and Example Activities Assessment • Identification of pilot areas and population alongside individual risk issues that would impact on objective evaluation of the pilot e.g. pre-existing social, economic, health conditions. • Identification of existing support mechanisms in pilot areas. • Stakeholder engagement including raising awareness within the public and voluntary sector and with other stakeholders about the basic income pilot. • Further community capacity and engagement work to assess current social networks and support, including employer engagement. Recruitment • Stage of recruiting participants within identified pilot communities for basic income. It is considered that consent will be required from participants to allow the collection and sharing of data over the period of the pilot and evaluation. • Awareness raising of participation in basic income pilot. • Establishing points of contact in each pilot area for participant questions and concerns. • Invitation to be part of the pilot population. • Participant management process and database. • Provision of clear inclusive information for prospective participants. • Training for community champions and workers to support recruitment and participation in the pilot. Pre-pilot Support • Following successful recruitment, support work with participants to ensure an equitable opportunity to participate in the pilot. • Ensuring appropriate CBI payment mechanisms in place for individuals (including children). • Checking current benefit entitlements undertaken by existing advice agencies. Transition ON • Commencement of pilots at defined date and delivery as per the characteristics of the basic income pilot. • Agreement for participating in the pilot.

101 Stage Definition and Example Activities Pilot Phase • Access to existing advice and guidance agencies for the provision of e.g. • Budgeting and financial support to participants • Training and skills development • Benefits and welfare advice Transition OFF • End of pilot at defined date and implementation of post-pilot support or • Decision by participant to stop participation in the pilot and leave the pilot population or move out of the pilot area or due to death of a participant. • Early ending of pilot. • Tailored support to exit the pilot. • Reassess financial and inclusion support and entitlements. Post-pilot Support • Support for participants following the end of the pilot. • Retaining and wraparound services in pilot areas. Change of Circumstance Data sharing agreement and shared approach with relevant local and national agencies to recording and managing change of personal circumstance for participants across the pilot. This will work at the following levels: • The Pilot – recording change in circumstance as it relates to the delivery and evaluation of the pilot including personal information; • Tax and Benefits – management of the change of circumstance for tax and benefits purposes with DWP, HMRC and Scottish Social Security to manage: − change in entitlement to benefits − change in personal circumstances affecting tax and benefits − changes in benefits.

102 7.3 Assessing and minimising the potential impacts on different groups

To assess and understand the potential impacts of a CBI pilot on different groups of people, the Steering Group conducted an indicative Integrated Impact Assessment (IIA). The IIA template used was based on guidance for councils, Health & Social Care Partnerships, Health Services and other relevant public sector bodies. It helps public bodies meet their legal duties to consider equality, human rights, socio-economic disadvantage, sustainability and the environment in planning and policy decisions. It is based on guidance produced by Lothian local authorities and NHS Lothian.xlix

The work undertaken does not include a full or final IIA report. For the purposes of feasibility assessment, an indicative assessment was deemed sufficient to identify the potential impacts of a CBI pilot. Should a CBI pilot go ahead in Scotland in the future, it is recommended that the agreed pilot model is assessed in full using the comprehensive IIA process.

The indicative IIA comprises a checklist of key questions and areas of focus to help identify what might change as a result of the pilot proposal and the differential impacts (both positive and negative) on different groups of people. It is based on the pilot model, including both lower and higher CBI payments, as described in Chapter 5.

A saturation CBI pilot is intended to include all individuals within a specified geographical area. It could therefore include individuals from a variety of backgrounds and circumstances with a range of characteristics. The intended principle is that CBI brings a range of potential positive impacts to different groups of people. However due to the complexity of interaction with current tax and benefit systems as well as other policy areas, there may be groups who are at a disadvantage compared other groups. The checklist and a list of key evidence sources is contained in full at the end of the feasibility report in Appendix 5. A summary of the notable impacts in relation to equality and human rights, differential impacts on different groups of people and an indication of key groups who may be at risk during a pilot of CBI is provided below.

There may be a range of positive impacts in relation to equality and human rights for people involved in a pilot. This includes reducing poverty and advancing the equality of opportunity by reducing social or income related barriers (e.g. lack of secure, regular or independent income) which may normally prevent people from accessing learning, training, employment, entrepreneurship, volunteering and social or leisure activities. It may enable people to have more control over their social / work environment by promoting participation, inclusion and a sense of dignity and control over decisions.

Some negative impacts may include pilot participants at risk of harassment from people not involved in a pilot who consider their exclusion to be unfair or unjust. Similarly, participants may be exposed to criticism based on the perception that they are being given ‘free money’. Those receiving the high-level CBI may be at particular risk of being exposed to a negative response from others.

In relation to differential impacts across people with protected characteristics, it is anticipated there would be a greater number of positive benefits than negative impacts. However, where evidence is mixed, or there is uncertainty over the potential impacts, key groups could be at risk of differential disadvantage during a pilot of CBI. These include: xlix Integrated Impact Assessment Guidance produced by Lothian local authorities and NHS Lothian can be found here: https://www.nhslothian.scot/YourRights/EqualityDiversity/IADocuments/ IntegratedImpactAssessmentGuidance.pdf

103 a. People vulnerable to falling into poverty – People in receipt of welfare benefits

There are premiums / top-ups associated with some welfare benefits which are available to those with additional needs related to disability, limited capability for work, caring and childcare. If these additional elements were not retained there would be a risk of absolute financial detriment and significant relative detriment for these individuals.

There is a risk that due to suspension of some entitlements, participants would lose access to a range of passported services and welfare benefits. The Steering Group recommends entitlement to passported services and benefits is retained. If, however, this is not confirmed through regulations, there is a risk that participants would lose access to Jobcentre Plus support services, National Insurance Credits and passported benefits resulting in financial detriment.

If some benefits are suspended, there is a risk of participants permanently losing out on transitional protection due to a pilot. Participants would therefore be at risk of relative and actual detriment compared to those not involved in a pilot. Some people may be unwilling to participate if they are automatically moved to Universal Credit upon finishing a pilot as some may be worse off financially.

People on benefits at greatest risk of adverse impacts include people on low incomes or with few other sources of income, people or families with complex needs and people whose circumstances change substantially during the course of a pilot. b. People vulnerable to falling into poverty – All groups

Generally a CBI seeks to reduce poverty and deliver positive impacts for all individuals who are at risk of falling into poverty e.g. removing the risk of destitution, reducing stress and stigma associated with being on welfare benefits, providing a stable income, improving the transition into employment and reducing the risk of losing income if unable to meet various conditions.

However, it will be important to ensure that existing or additional support for vulnerable families can be accessed throughout and beyond a CBI pilot. An individual CBI may support people to leave unsafe situations (if this was the case and they moved outside a pilot area then there would need to be safeguards to ensure the money would still follow them). There is a risk that receipt of CBI (particularly at a high level) could put some vulnerable individuals at risk of financial exploitation. A process to identify these individuals and provide support will be necessary. There is risk of negative impacts associated with a sudden drop in income when CBI stops at the end of a pilot, requiring careful consideration of a safe and possibly phased exit transition. Depending on the design of a CBI pilot model, there could be a risk of financial and other forms of detriment for people on low incomes. Section 7.2.3.1 provides further detail on these risks and possible mitigating solutions. c. Disability – People with disabilities (including physical disability, learning disability, sensory impairment, long-term medical conditions, mental health problems)

Relative poverty rates are higher for people with disabilities than for those without. In this regard a CBI, particularly at the higher level, is expected to have a positive impact on people with disabilities who are facing poverty. People with disabilities also have higher living costs than those without disabilities. To ensure there is no relative financial detriment for people with disabilities and they have funds to meet these additional needs, it is proposed within the pilot model that all entitlements and elements relating to disability and work capability (also including housing and childcare support), is maintained for the pilot duration. If these entitlements and elements can be maintained,

104 people with disabilities should not face a negative impact. However due to the complexity of Universal Credit and legacy benefit systems there are risks associated with maintaining these entitlements and elements alongside a CBI pilot. d. Age – Older people of pension age

Due to the suspension of State Pension during a pilot, older people would likely be asked to defer entitlement to their pension. State Pension can only be deferred once during a lifetime. Participants would therefore lose the opportunity to defer their pension entitlement at a later date.

For some policy areas or groups of people, a lack of available or relevant evidence makes it difficult to understand the potential impacts of a pilot. In such cases additional research or public engagement is recommended to fill these evidence gaps and ensure an informed judgement can be taken. In other cases, particularly in relation to people with disabilities, upon agreement of a preferred pilot model, further engagement or evidence may be required to understand the intricacies of potential impacts for people with disabilities.

The following groups or policy areas are identified as requiring further research or engagement in relation to the potential impacts of a specified CBI pilot model:

• Minority ethnic groups • People with disabilities • Marriage and civil partnership • Pregnancy and maternity • LGBTQI groups • Religion / belief groups • Supporting local businesses • The environment and sustainability

The design features of a pilot model will determine how some groups are affected e.g. the interaction of CBI with benefits and tax could impact on people with disabilities, marriage and civil partnership and pregnancy and maternity. We therefore recommend that any agreed CBI pilot model is carefully considered in relation to the potential impacts on equality and human rights and people with protected characteristics using the comprehensive Integrated Impact Assessment process.

7.4 Protection of participants’ rights: Participation and informed consent

7.4.1 Participation in a policy pilot A fundamental principle of social research is that participation should be based on informed consent. In the context of a clinical trial for a new drug treatment, for example, potential participants would be provided with information about the study to enable them to make an informed (and voluntary) decision whether or not they wish to participate. Subsequent allocation to the intervention or control ‘arms’ by those consenting would be done at random. Evaluating the effectiveness of a pilot policy based on delivery of an intervention (a CBI) to a whole community within a saturation site selected at random, however, may pose policy governance and research ethics questions. In particular, it requires consideration of:

105 • The grounds for participation in the policy pilot • The requirement to obtain informed consent to participate in an evaluation of the policy pilot • The implications for obtaining informed consent from people who are members of communities which may already have been randomly allocated to intervention ‘saturation sites’ • Grounds for participation in a policy pilot.

There is an argument that participation in a policy pilot may not always require to be voluntary. The Social Security Advisory Committee (SSAC), for example, which examined ethical and legal issues of piloting in the Department for Work and Pensions (DWP) argues that: ‘it is acceptable for a number of individuals to be mandated to participate in the pilot so that either the expected benefits can be rolled out to the rest of the eligible population or the policy can be altered or rejected at an early stage’.59 According to the SSAC, foregoing voluntary participation is justified by the fact that the government is required to deliver the best possible conditions for all of its subjects. The SSAC states that ‘some degree of risk within pilots may be defended on the grounds that it will produce information that is likely to increase the “common good”’. As noted above in relation to processes of inclusion and exclusion, mandating people to participate in a policy pilot would only be permissible insofar as the policy is within the legal competence of policy makers, and that data collection does not infringe participants’ human rights.60

A key argument for compulsory participation in a policy pilot is that, in principle at least, it would remove selection bias arising from voluntary participation, with implications for assessing effectiveness, and therefore the robustness of any evaluation. In Finland for example, participation in the CBI experiment was compulsory for unemployed individuals selected at random.61 Compulsory participation also provides an opportunity for learning about the administration of a new system. In addition to providing information on the reach of the CBI (i.e. who actually receives it), the collection of administrative data would also enable an analysis of some of the outcomes in the sample population. It should, however, be noted, that even where participation in a policy pilot is compulsory, participation in an evaluation that involves the collection of personal data would still require the informed consent of potential participants.59 As noted in Section 6.3.2, this has implications for response rates and therefore the sample size required to measure changes in individual level outcomes. Compulsory participation in the policy pilot does not therefore necessarily remove the problem of selection bias.

In relation to grounds for participation in a policy pilot, the Steering Group have identified three options. Each of these has a number of implications for the implementation of the policy pilot, the quality of an evaluation and research ethics. a) Participants opt-in to a policy pilot

This may be considered the most ethically sound option insofar as it allows participants to actively volunteer to participate in a policy pilot, i.e. to receive a CBI. However, this may mean that the pilot recruits insufficient numbers to ensure robust quantitative evaluation, even if all were to consent to participate in an evaluation of the pilot. There is also a risk of selection bias where some groups may be under- or over-represented across the sample. There would also be implications for understanding community level effects if, in practice, only some members of a community chose to participate.

106 b) Participants able to opt-out of a policy pilot

This option has similar risks and implications for the quality of an evaluation of a pilot as outlined in option A above.

There may be a valid rationale to offer this opt-out option if the delivery team are concerned that the risks or issues of transitioning on or off a pilot, or changes in circumstances in the course of a pilot cannot be adequately mitigated. This may be particularly relevant to participants who would be transitioning to or from welfare benefits and there are concerns about fluctuations in income and the availability of transitional benefit protection. c) Compulsory Participation in a policy pilot

The operating principle of the feasibility study is avoidance of detriment to participants. As long as this principle is realised in practice, it may be acceptable to mandate individuals to participate in the pilot (but, as noted above, informed consent would still be required for the collection of personal data for evaluation purposes). If CBI is to work as a benefit replacement, individuals may be mandated to participate insofar as this replacement does not cause any detriment to them. Otherwise, they could be mandated to participate only if they would be able to retain access to all existing benefits and possibly have CBI income disregarded for the purpose of assessing entitlement to them, thus avoiding detriment.

It should also be noted that, compulsion may have different implications for different groups. For people on welfare benefits, a view has been expressed that if (existing) welfare entitlements were suspended then receipt of a CBI could be effectively compulsory for those without other financial means. The differential distribution of potential risks and harms may have implications for policy governance as well as research ethics.

To maximise learning we recommend that if a pilot were to go ahead, that consideration is given to compulsory participation in the intervention, but that the Scottish Government should also seek expert advice on the legal and procedural basis for requiring people to accept a CBI. This again would need to be based on the model of CBI that would be piloted.

7.4.2 Informed consent to the collection of personal data Obtaining informed consent is a fundamental principle of social and medical research: consent has to be informed, freely given, and can be withdrawn at any time without reprisal (with appropriate consent procedures for children under 16 years, and for adults who may lack capacity).l, 62, 63, 64 As such, even if it is within the legal competence of policy makers to mandate participation in a policy pilot, participants would still have the option to withhold consent to providing personal information for research and evaluation purposes. In the context of the suggested cluster control evaluation design (see Section 6.3), where all individuals living in whole communities selected at random comprise the intervention group, this raises three specific issues:

• The implications for individual informed consent where the CBI is delivered at the individual level, but randomization occurs at the community level65 l The UK Government Social Research Unit’s (2011) Professional Guidance for Ethical Assurance for Social Research in Government, states that ‘Participants in any research study involving primary data collection must be asked for their consent to take part unless the law requires participation, as with the Census.’ (p.9) https://assets.publishing.service.gov.uk/government/uploads/system/uploads/ attachment_data/file/515296/ethics_guidance_tcm6-5782.pdf (accessed 26 February 2020)

107 • Informed consent to data collection • The implications of non-response or consent withheld to participate in the evaluation.

7.4.2.1 Informed consent and Cluster Randomized Trials (CRTs)

Because cluster randomized trials randomize at the social group level e.g. a community, rather than at the individual participant level, they raise particular issues. For example, clusters, or, in the case of a pilot CBI, communities, may be randomized before individuals can be identified and recruited and consent to their participation in the pilot obtained. Similarly, if the cluster is a whole community it may be difficult or impossible for an individual to avoid, precluding meaningful refusal to participate.66 In these contexts the Ottawa statement on the ethical design and conduct of cluster randomized trials sets out a set of recommendations.66 In relation to informed consent these include:

• Researchers must obtain informed consent from research participants, unless a waiver is obtained from a Research Ethics Committee (REC) under specific circumstances. Informed consent refers to randomization, study interventions and data collection procedures • In CRTs where identification and recruitment are not possible before randomization of clusters, participants can be enrolled after randomization. Consent should, however, be sought as soon as possible after potential participants have been identified and before administering the intervention or starting data collection • CRTs may contain vulnerable participants. For these groups researchers and RECs must consider whether additional protections are needed. • Where individuals in a cluster are less able to choose participation freely because of their position in a cluster, special attention should be paid to recruitment, privacy and consent procedures for these participants. This may be relevant in the context of a pilot CBI where, for example, those in receipt of welfare benefits may feel less able to freely choose to participate in an evaluation of the pilot. • A REC may approve a waiver or an alteration of consent, if the research is not feasible without a waiver or alteration, and the study and data collection pose no more than minimal risk to participants. If informed consent is possible for some but not all the study interventions or data collection, informed consent should be obtained, where possible, for each procedure. For example, in a CRT involving a public health intervention where a REC has approved a waiver of consent to the study intervention (e.g. receipt of a CBI), informed consent to data collection procedures may still be required.

We recommend that, prior to the commencement of a CRT of a pilot CBI, approval is obtained by a REC. It is likely that this would be an ethics submission made by the researchers undertaking the evaluation. Any submission would be accompanied by the evaluation protocol, and would include the rationale for the CRT, as well as information on procedures for obtaining consent, and data collection materials, including consideration of the additional protections for vulnerable groups, or those within a cluster who may be less able to freely choose to participate because of their position in the cluster. While a REC could be a multi-site Medical Research Ethics Committee (MREC), given that the proposal would be for an evaluation of a non-health intervention, a university ethics committee, or an equivalent to the NHS Health Scotland Research Development Group may be more appropriate.

108 7.4.2.2 Informed consent to data collection procedures

As discussed in Section 6, there are a number of potential sources of process and outcomes data. For many of the outcomes (and processes) of interest ‘bespoke’ data would need to be collected from individual participants. This could be, for example, by survey, in-depth interview or focus group. Informed consent would be required for these data collection processes.

Data on some of the outcomes of interest could, however, be obtained from routine data collection, whether collected as part of ‘usual practice’, for example, by DWP or HMRC, or collected as part of the delivery of the CBI. This would have the advantage of reducing the burden and risks of data collection on participants. However, data must to be anonymised prior to transfer for use for research purposes. There may be scope for data linkage to assess, for example, the relationships between income/receipt of a CBI and health, employment, education and training outcomes. This would require data sharing agreements between different agencies. For data originated in NHS Scotland, approval for data linkage would be required from the Public Benefit and Privacy Panel (PBPP).li

The Scottish Government Guiding Principles for Data Linkage (published in 2012) suggest that, where practicable, consent should be obtained from data subjects prior to linkage of personal data. This can only be departed from where there is a strong justification for not obtaining consent and approval has been obtained from an appropriate approval body (such as, where appropriate, the PBPP). This body could confirm that the proposed data linkage was in the public interest, and that the appropriate safeguards were in place to minimise the risks of identifying (or re-identifying) individuals.67

As noted above, data collection and informed consent procedures will need to take into account that a community-based cluster will include babies/young children, adults with incapacity, and other vulnerable groups.

It is recommended that informed consent is obtained for all new/bespoke data collection for evaluation purposes, and for data linkage between new/bespoke data and routinely collected administrative data. It is suggested that appropriately anonymised routinely collected administrative data could be used for research purposes without obtaining individual consent (subject to the appropriate information governance and data protection procedures and protocols of the relevant agencies). We recommend, however, that independent advice is obtained to confirm the appropriateness of accessing these data without consent. We also recommend that Scottish Government undertake a Data Protection Impact Assessment, prior to piloting and evaluating a pilot CBI.

7.4.2.3 Consent withheld/non-participation

Even if compulsory participation in a policy pilot is deemed lawful, the requirement to obtain informed consent to the collection of ‘new’/bespoke data raises the likelihood of withheld consent or non-participation in the evaluation. As discussed in section 6.3.3 above, this would have implications for the sample size achieved (and therefore the power to detect changes in the primary outcomes of interest). If some groups are more or less likely than others to ‘opt out’ this may also create a biased sample. This applies to both the intervention and control groups. The risks of non- participation in an evaluation may, in fact, be even greater in the control group who have less stake li See https://www.informationgovernance.scot.nhs.uk/pbpphsc/ for more information.

109 in participating in an evaluation of an intervention they do not receive. To address this, two options have been considered:

• Over-sampling. This would have implications for the size of the saturation sites to take into account non-response. In the intervention sites this would also increase the cost of the intervention (more people would need to receive the CBI than necessary for the evaluation) (see Section 6.3.3). • ‘Compensating’/incentivising people to participate. Providing a financial incentive to encourage participation was considered but is not recommended on three grounds. First, given that CBI is a form of cash transfer, giving people an additional (financial) incentive to participate would potentially distort or contaminate the findings; second, contingent on the amount paid, from a research ethics point of view it could be considered to be ‘unwarranted material gain;lii and, third, it could be viewed as a form of coercion, rather than a gift acknowledging participation.

We recommend that to address nonparticipation in an evaluation of a pilot CBI, consideration is given to over-sampling to achieve statistical power.

7.4.3 Routes for identifying and recruiting evaluation participants As noted above, if the target population is a saturation site(s) all those within the site would effectively be a (potential) research participant. Nonetheless, for the purposes of including people in the study, a mechanism needs to exist for identifying and approaching potential participants in the first instance. If this is through an agency charged with delivering the CBI this could raise participant concerns about privacy and confidentiality. As part of whatever process is involved in ‘signing up’ for a CBI, consideration would also need to be given to obtaining consent to contact for evaluation purposes (consent to participate would be part of a separate process undertaken by the evaluation team). Consent to the use of data would also need to be considered at that stage.

If administration of a CBI is handled by an organisation already administrating welfare benefits payments or tax returns e.g. DWP, HMRC, Social Security Scotland, it is recommended, that consideration is given to processes for marking cases in receipt of a CBI. Whether administration is by a pre-existing agency or an agency set up with a specific remit to administer CBI, information governance protocols and mechanisms would need to be in place to enable agencies to obtain the consent of recipients to their details being passed on to an evaluation team. Anonymised information would also need to be obtained to indicate the numbers and characteristics of those approached, so that (non-) response rates can be monitored. The potential for learning from evaluations of Universal Credit (https://revenuebenefits.org.uk/universal-credit/policy/evalutaion- of-universal-credit/) and from the measurement framework for the Scottish Social Security Charter (http://www.gov.scot/ISBN/9781839602474), when implemented, could be explored, once the mechanism for administering a pilot CBI was agreed.

7.5 Governance recommendations

It is recommended that, if a policy pilot is progressed, and a Cluster RCT proposed for an evaluation of the policy pilot that:

lii The RESPECT Code of Practice for socio-economic research includes the principle that researchers should ensure that the research process does not involve unwarranted material gain or loss for any participant. http://www.respectproject.org/code/respect_code.pdf. (Accessed 23 July 2019)

110 a) prior to the commencement of an evaluation ethical approval is obtained from an appropriate Research Ethics Committee b) appropriate legal and procedural advice is sought to ensure that the selection of intervention areas (and exclusion of other areas), and the mandating of people to participate in a pilot is within the legal competence of the policy making body, procedurally fair and reasonable. c) participants, particularly those who are vulnerable and/or on low incomes, should not experience detriment (financial or otherwise) compared to individuals not involved in the study. The risk of detriment can be reduced by ensuring CBI payments are disregarded as income for means-tested benefit calculations. d) the agreed pilot model is assessed in full using the comprehensive IIA process. e) to address non-participation consideration is given to over-sampling to achieve statistical power. f) Scottish Government undertakes a Data Protection Impact Assessment. g) for data linkage, data sharing agreements are put in place between relevant agencies at an early stage and a submission is made to the PBPP. h) early consideration is given in the development and design of the CBI delivery mechanism, to the processes for identifying CBI recipients where the agency covers more than CBI, and for appropriately obtaining consent to contact for the purposes of evaluation.

111 Section 8: Feasibility Assessment

8.1 Strategic Feasibility

8.1.1 Strategic Feasibility Concept Strategic feasibility relates to the strategic action to build a robust political coalition of support for enabling the legislation and subsequent adoption of CBI as public policy. This may involve politicians, political parties, social movements, interest groups, trade unions or other organised groups. Clearly different groups have varying levels of power and resource available to influence the policy making process, but it is not just the degree of their support that is of interest. Different groups may support different models of CBI, or have varying outcomes in mind for the policy, thus strategic feasibility is not straightforward and depends on the proposed CBI model and implementation levers required.

8.1.2 Strategic Landscape 8.1.2.1 Governmental and Political Positions

The UK General Election in December 2019 highlighted the differing views across the political spectrum in the degree of support and opposition to basic income. The 2019 Green Party Manifesto incorporated a pledge to “… the creation of a Universal Basic Income, paid to all UK residents to tackle poverty and give financial security to everyone”.liii A report40 by Professor Guy Standing in 2019 for the then Shadow Chancellor of the Exchequer John McDonnell, resulted in the 2019 Labour Manifesto pledging to “…explore other innovative ways of responding to low pay, including a pilot of Universal Basic Income”.liv Prior to the 2019 UK general election, a number of Liberal Democrat candidates signed an open letter in support of trialling an unconditional minimum income element in the social security system.lv This demonstrates that although interest exists across the UK political spectrum, there is no consensus on the correct model for a basic income or its relationship with the existing welfare state.

UK and Scottish Parliamentary committees have taken evidence and discussed basic income but again expressed mixed views on the efficacy of the policy and on whether a pilot is to be supported even from afar. The Scottish Parliament’s Social Security Committee held an oral evidence session in March 2017 with input from the Centre for Research in Social Policy, Loughborough University; Reform Scotland; Citizens’ Income Trust and Trustee of Citizens’ Basic Income Network in Scotland; Landman Economics and the RSA. The Committee also received an update from the Steering Group and had an opportunity to ask questions on the Feasibility Study in November 2019.

The Scottish Government have supplied a key supporting role to the Steering Group and engagement with civil servants and Scottish Ministers has allowed constructive dialogue to take place throughout the development of the feasibility study. The Steering Group has engaged with the Department of Work and Pensions and Her Majesty’s Revenue and Customs around the feasibility work for a CBI pilot. The insight and depth of knowledge has allowed the Steering Group to assess the feasibility of its proposals for pilot models based on the current legislation liii 2019 Green Party Manifesto https://www.greenparty.org.uk/assets/files/Elections/Green%20Party%20 Manifesto%202019.pdf liv 2019 Labour Party Manifesto https://labour.org.uk/manifesto/tackle-poverty-and-inequality/ lv https://thoughtsofprogress.wordpress.com/2019/11/10/lib-dems-backing-minimum-income-pilots/

112 and processes. This has highlighted the probable need for primary UK legislation to allow a pilot of the kind proposed to be undertaken in Scotland. Although some benefit and welfare powers have been devolved to the Scottish Government, these are insufficient to allow a pilot in Scotland without specific primary legislation.

8.1.2.2 Cross-Party Group in the Scottish Parliament on Basic Income

The Cross Party Group (CPG) on Basic Income is a Scottish Parliament working group. It was established independently of the Steering Group and had its first official meeting on 20 June 2018. Approval to establish the group was given by the Scottish Parliament’s Standards Committee on 7 June 2018. The group is currently co-chaired by Tom Arthur MSP and Alex Rowley MSP. Secretariat for the group has been provided by the RSA and Citizens Basic Income Network Scotland. The Steering Group presented to the CPG in November 2018 and provided an update in November 2019.

8.1.2.3 Local authority interest in a basic income

Section 2.1 of the introduction to this report sets out the background to the four Scottish local authorities interest in exploring a CBI pilot. In March 2018, COSLA (Convention of Scottish Local Authorities) considered a paper on the Basic Income and the Feasibility Work being undertaken in Scotland. A further update was provided to COSLA following the launch of the Interim Report by the Steering Group in November 2019. COSLA continue to be supportive of the work of the Steering Group.

In addition to the feasibility work in Scotland, active pursuit of piloting basic income is being explored by local authorities in England and Wales. The most active are in the north of England, where some local authorities have taken part in fora and in some instances, passing motions in support of the policy.lvi

8.1.2.4 The role of civil society

The role of civil society and of those organisations that develop and inform opinion has proven crucial in developing and maintaining momentum in relation to basic income and the need for piloting the policy within Scotland.

In 2015 the RSA released the first of a series of reports that explored the role and impact of a basic income. A number of other reports also were released that explored the need, benefits and issues of a basic income within the UK. These reports and basic income schemes are identified in Section 4.3 of this report.

The Basic Income Earth Network (BIEN) is the international body leading the campaign and sharing learning and developments on basic income. Its UK counterpart the Citizens Basic Income Trust and Scottish group Citizens Basic Income Network Scotland (CBINS) have developed campaign and educational material. In 2018 CBINS, funded by the Scottish Universities Insight Institute, led a series of cross-disciplinary workshops as part of the ‘Exploring Basic Income in Scotland’ project. The project brought together a range of academics, policy-makers, organisations and citizens to explore issues and potential impacts of basic income on a number of policy areas.lvii These lvi Councils in Liverpool, Hull and Sheffield have all expressed an interest in hosting pilots of basic income. lvii A full report on the project, and a series of individual reports focussed on specific policy areas are available at: https://www.scottishinsight.ac.uk/Programmes/Scotland2030/BasicIncome.aspx

113 areas included employment and entrepreneurship, housing, care, human rights and equality, and modelling, implementation and evaluation.

Many of these policy development and influencer groups are positive about the concept of basic income and supportive of the piloting of the approach. However, a large number of groups within the field of inequalities and poverty are at best agnostic to the policy and are against the introduction of this approach and/or against the implementation of a pilot. As part of Stakeholder engagement sessions, these groups have engaged with the feasibility study and have been actively encouraged to share concerns to ensure a robust discussion of the issues needing to be addressed.

In particular, the debate about whether a basic income would have a positive or negative impact on the lives of disabled people is the subject of much discussion. The Steering Group are also aware of the need for a basic income to interact with some existing benefits and entitlements in a way that causes no detriment to the financial well-being of disabled people. At the end of July 2019 Inclusion Scotland, a national network of Disabled People’s Organisations and individual disabled people, produced a briefing and organised a ‘pop up think tank’ to discuss the potential impact of a basic income on disabled people. The briefing highlights the additional financial costs facing disabled people and promotes a discussion about the benefits or otherwise of basic income. Inclusion Scotland are keen that disabled people are adequately considered as a basic income policy develops in Scotland. A report in early 2020 by Inclusion Scotland laid out the concerns of the group in relation to the implementation of a basic income.68

Trades Unions have expressed mixed views regarding basic income. The Trades Unions Congress (TUC) motion in 2016 supported basic income as complementary to comprehensive public services and childcare provision and proposed that it should include supplementary benefits to support people on low incomes with high housing costs, and disabled people. The motion continued by noting that any transitional arrangements should leave people with lower incomes better off. It also proposed that the basic income should work in tandem with strong trades unions and employment rights and the provision of secure and properly paid work. Arguments from Trades Unions against basic income are that it can entrench low pay and precarious work and that it is unaffordable. There is concern the cost of a CBI would require reductions in other services that are required to deliver important social and economic goals and as such some do not consider it an effective route to reforming welfare, arguing that it diverts effort from more pressing and progressive issues that should be at the core of the Trades Unions Movement. Others also consider it to be an individualist solution to a shared set of problems. In summary, the views of Trades Unions range from those that see it as a utopian remedy that does not address core issues, through to those that believe that proponents of basic income and Trades Unions have much in common and should exploit this to address a changing labour market. There are also views expressing support for differing models of basic income.

8.1.3 Alternatives to Basic Income Across the range of governmental and non-governmental organisations exploring basic income, a range of alternatives have been proposed. These have included:

• Negative Income Tax • Job Guarantees • Reducing hours of paid work

114 • More and better quality public services • Minimum and living wages • Social Insurance • Means-tested social assistance • Subsidies and vouchers • Workfare • Participation Income • Universal Basic Services • Workfare • Tax credits • Charity

Arguments in support or against these are many and out with the scope of this feasibility study although we have considered other baseline scenarios in the modelling of our approach to piloting basic income.

8.1.4 Strategic Challenges Strategic feasibility includes determining whether, across different interest groups, and between different levels of government, there is political support for, and consensus can be reached around, CBI pilots.

Basic income is an idea that appeals across the political spectrum, but often for different reasons, and a different underlying rationale. While there is a broad agreement on the key principles of a basic income, different political choices may lead to very different types of basic income schemes being proposed.

Our International Learning Report14 identified two key political and strategic considerations for a CBI pilot: a. Connecting constituencies of support

The level of government participation is likely to be a critical success factor. In the case of Scotland, collaboration is required with both Scottish and UK governments if piloting basic income is to be considered feasible. b. Understanding the political cycle

Political events have the potential to shape the design and future direction of pilots. In July 2018 it was announced that the Ontario experiment would be terminated early. This sudden decision demonstrates the risk that changes in political actors may change the level of commitment to a pilot study, particularly if a pilot were run over several years and coincided with elections.

8.1.6 Conclusions This section has provided a summary of the key actors in the strategic landscape and political challenges associated with delivering a pilot CBI. Piloting a basic income will require political support that can be sustained over time. Interest from the Scottish Government and clear support

115 in two national party manifestos shows that there is a continuing interest in exploring how basic income can tackle inequalities and precarious income. The need for primary legislation illustrates the complex institutional issues that a pilot will face and the legislative time that would be required to draft and pass UK Parliamentary Legislation within a full schedule and other priorities of the UK Government. Options for addressing this issue include:

• Continuing engagement between Local, Scottish and UK Governments in developing the necessary frameworks and legislation for piloting basic income; • A private members bill to establish the basic income pilot in Scotland; • Exploring other options for piloting aspects of basic income that can be addressed through existing regulations and powers available at local, Scottish and UK levels.

8.2 Institutional Feasibility

Institutional feasibility is one aspect of feasibility which the Steering Group has explored in detail. This includes determining whether there is institutional commitment from a range of organisations (including the Department for Work and Pensions (DWP) and Her Majesty’s Revenue and Customs (HMRC)) to support the implementation, administration and funding of CBI pilots. To fully explore the institutional feasibility of a CBI pilot it is important for the Steering Group to understand the potential interactions between a CBI and tax and welfare benefits systems.

8.2.1 Engagement with Relevant Institutions The Steering Group have engaged with the DWP since April 2018. In an exchange of letters between the former Cabinet Secretary for Communities, Social Security and Equalities Angela Constance MSP and the then former Secretary of State for Work and Pensions, Esther McVey MP, the then Secretary of State committed the Department for Work and Pensions (DWP) officials to engage with the Scottish Government and local authorities on the project (letter dated 19th May 2018). In November 2019, the current Cabinet Secretary for Communities and Local Government, Aileen Campbell MSP, wrote to Thérèse Coffey MP the current Secretary of State for Work and Pensions and Jesse Norman current Financial Secretary to the Treasury, to seek assurance officials would continue to liaise with the Steering Group. Face to face meetings have been held with DWP and HMRC in March 2019 and January 2020, which have highlighted that more detailed discussions are necessary to fully map out the best way forward with benefit and tax interaction.

Any Scottish basic income pilot will need to rely on the full collaboration of DWP and HMRC and it is recommended that before any pilot is considered, further detailed discussion takes place with the DWP and HMRC. In addition, the issue of how a CBI would interact with the complex range of entitlements that those of pension age currently receive requires careful consideration. The meeting in January did not have the opportunity to discuss how state pension policy would interact with a CBI pilot so further engagement is recommended to ensure this is satisfactorily explored.

8.2.2 Welfare Benefits-CBI Interaction Research The Child Poverty Action Group in Scotland (CPAG) were commissioned in January 2019 to consider how a pilot study of CBI might impact on pilot participants’ eligibility for other welfare benefits and associated ‘passported’ benefits. The research, published in June 2019, highlighted the complexities of the current social security systems and the challenges involved in putting a basic income pilot in place.69

116 The main aims of the research were:

1. To gain an understanding of the benefits and tax implications for different population groups for different levels of CBI; 2. Provide analysis of the relevant social security and tax legislative and policy constraints and opportunities at UK, Scotland and local authority levels for delivering a (no detriment) CBI.

Research concluded that leaving all benefits in place and ensuring CBI income is disregarded for calculation of means-tested benefits is the surest way to avoid detriment to pilot participants. However, such an approach may be of limited value in learning about the effects of a CBI scheme which would replace parts of the current benefit system. The Steering Group have used the research to inform the design of a pilot which attempts to reduce the risk of detriment to participants while also providing scope for learning about the effects of a CBI. Further information is provided in section 7.

8.2.3 Legal considerations The Steering Group approached SOLAR (Society of Local Authority Lawyers and Administrators in Scotland) for informal feedback on the ability of Scottish local authorities to pilot a basic income within existing statutory powers, regulations and guidance. While informal feedback was shared, SOLAR suggested that full independent legal analysis on a national level, rather than by individual local authorities would be required. Additionally, in July 2019 Scottish Government policy representatives participating in the group sought advice from Scottish Government lawyers on the proposed models to support further discussion between Scottish Government and both DWP and HMRC.

8.2.4 Key Principles of a CBI Pilot To adequately test the effects of a CBI, the Steering Group has adhered as closely as possible to the overall principles of CBI: periodic; in cash; individual payment; universal; and unconditional.

A key principle of the CBI feasibility study is that pilot participants (particularly vulnerable and low- income groups) are not financially worse-off as a result of participating in the study (either during the pilot or beyond). To this end, the Steering Group emphasise the importance of retaining welfare support for the greater needs of some groups, specifically participants with disabilities, limited capability for work, carers, housing and childcare needs.

8.2.5 Delivery Mechanisms There are various legislative powers and delivery mechanisms to be assessed in the context of a CBI pilot. The following section identifies these options, providing a summary of the key constraints and implications of each delivery mechanism.

8.2.5.1 Delivery Powers

The current social security system in Scotland is administered across three different levels of government: UK Government (DWP or HMRC), Scottish Government via Social Security Scotland, and Local Authorities who deliver benefits under rules set by both UK and Scottish governments. UK Government, Scottish Government and local authorities all have various existing powers which could be mechanised to pay a Citizens’ basic income and deliver a pilot in Scotland. It is also noted that a pilot CBI could potentially be delivered via a charitable trust or private company.

117 Table 15 summarises the powers which could be used to deliver a CBI. It also provides an assessment of each option, identifying key constraints or implications for the feasibility of a CBI pilot model as specified by the Steering Group.

118 Table 15: Feasibility Assessment of CBI Delivery Mechanisms

Delivery Mechanism Description Feasibility for CBI Pilot UK Government Undertaken by UK • No constraints on UK parliament to legislate for amendments to means-tested benefits Government on behalf which would support a CBI pilot. However, it is acknowledged that this may require of Scottish Government. considerable and time-consuming changes to existing legislation and IT systems. Potentially delivered by • UK Government would need to agree to undertake and deliver on behalf of Scottish DWP or HMRC Government. Scottish Government Permits ‘top up’ of a • Scottish Ministers permitted to top up any reserved benefit. - Exception 5 of the benefit reserved to UK • Under current rules, a prerequisite is for recipients to require financial assistance in Scotland Act 1998 Government. addition to the benefits they receive. • Under current rules this option restricts delivery to individuals entitled to social security benefits. • Further investigation required to determine whether it would be possible to incorporate those currently not receiving benefits into the payment system. • Changes in circumstances during a pilot may end an individual’s entitlement to benefits, and therefore force an early exit from a pilot. • There is a restriction on using this power to offset effects of conditionality in reserved benefits.

119 Delivery Mechanism Description Feasibility for CBI Pilot Scottish Parliament- Power to create new • Scottish Parliament has power to legislate for a new social security benefit, but only within Exception 10 of the benefits in areas of areas of devolved responsibility. This may restrict the design of a pilot using this delivery Scotland Act 1998 devolved responsibility. mechanism. • Primary legislation required to ensure clarity within complex social security legislation and ensure participants remain eligible for other benefits including housing, childcare costs and disability support. • Bar on creating new pensions which people qualify for on the basis of reaching pension age. This may have implications if eligibility for a higher amount of CBI is based on reaching pension age. • Restriction on using this power to offset effects of conditionality in reserved benefits. • Current Universal Credit and Legacy benefit regulations would likely treat CBI payments differently for purpose of calculating means-tested benefits. Local Authority - Power to enhance • Cannot use this power in a way which unreasonably duplicates an existing legal function, Statutory power to wellbeing of some or which may restrict the design of a pilot. However, this restriction does not apply if there is enhance wellbeing all people in an area, consent from legal body whose function it is. (Section 20 Local including providing • Likely to require consent and negotiation with both Scottish and UK governments. Government in Scotland financial support. Used Act 2003) in 2013 to establish the • In a review of Local Governance in 2018, SOLAR (Society of Local Authority Lawyers and interim Scottish Welfare Administrators in Scotland) noted the current power of wellbeing has been little used and Fund. its scope eroded and restricted by Court decisions. They contend that actions in reliance of the power are increasingly unsafe and the power now has little value.70 • Delivery via this mechanism could be of limited value for learning about the effects of a national CBI policy.

120 Delivery Mechanism Description Feasibility for CBI Pilot Local Authority - Used to deliver Scottish • These powers are restricted to cases of defined need. Discretionary payments Welfare Fund and • Would require primary legislation to amend terms of devolved legislation to align with to people in need provision of financial delivery requirements of a CBI. support to children and adults by social-work • Delivery via this mechanism could be of limited value for learning about the effects of a departments. Restricted national CBI policy. to cases of defined need. Non-Governmental Legislation may not be • May not require legislation to achieve. Delivery: Charitable required to pay via this • If funding is from Scottish or local government this may still be regarded by UK Trust / Private Company method. However, this is Government as encroaching on reserved social security. dependent on source of funding. • Therefore, will still require consent and negotiation with Scottish and UK Governments. • The treatment of payments from charitable trusts is unclear within the current benefit regulations for Universal Credit. Regulatory clarification would be required to ensure consistency of treatment. • Delivery of a pilot via this mechanism may be of limited value in learning about the impacts of a national CBI policy. • There may be a requirement for a data sharing agreement between a delivery charity and local authorities or DWP. General Data Protection Regulations may pose a significant challenge to sharing data.

121 The evidence outlined in Table 16 demonstrates the challenges of identifying a suitable mechanism to deliver a pilot that fully tests the main characteristics of a CBI. At this time, until the necessary changes are mapped out with DWP/HMRC and fully considered, none of the mechanisms could deliver a CBI pilot of the model proposed by the Steering Group. Each mechanism has significant restrictions and implications for the pilot model design. For example, other than delivery by UK Government or a Charitable Trust, it would not be possible to test a saturation type CBI where all individuals in a geographical area (whether in receipt of benefits or not) could receive a CBI payment.

In the event of it being possible to deliver a CBI pilot as defined in Chapter 5 using one of the powers identified above, there are substantial and additional complexities associated with doing so.

The treatment of a CBI payment for the calculation of means-tested benefits will have an influence on how much individuals would receive from these benefits. A CBI which is disregarded for means- tested benefit calculation would have no impact on the amount of money received from means- tested benefits. However, a CBI that was counted as income would reduce the value of the means- tested benefit and risk loss of entitlement to other passported support.

The delivery mechanisms have different implications for how a CBI might be regarded for the calculation of any retained means-tested benefits. Under current legislative rules, CPAG have indicated the likely ways in which CBI may affect entitlement to Universal Credit, Legacy benefits and Pension Credit (Table 16).

Table 16: Delivery Mechanism and CBI Consideration for Calculation of Means-Tested Benefits

Delivery Mechanism Universal Credit Legacy Benefits Pension Credit Scottish Government Disregarded in current Adult CBI counted, but Adult CBI may be - Exception 10 (new rules child CBI disregarded, counted as income in benefit) of Scotland Act as income in current current rules 1998 rules Local Authority - Disregarded in current Adult CBI counted, but Disregarded in current Statutory power to rules child CBI disregarded, rules enhance wellbeing as income Non-Governmental Rules unclear Disregarded in current Disregarded in current Delivery: Charitable rules rules Trust

In each of the three mechanisms identified above, CBI is treated differently for calculation of UC, Legacy benefits and Pension Credit. This would cause inconsistency and inequality in treatment across pilot participants and risk putting participants in financial detriment. Furthermore, in relation to the key principles of a CBI pilot, it is important that a preferred delivery mechanism and corresponding impact on social security entitlement does not undermine the delivery of an accurate test of CBI.

Following the evidence outlined above, at this time there is currently no clear preferred or feasible option to deliver a CBI pilot model as outlined in Chapter 5. All options would require substantial engagement and negotiation with UK, Scottish and local governments to avoid unintended consequences for pilot participants. This may take the form of time-consuming primary and secondary legislation to enable delivery of a satisfactory CBI pilot and there is a need to map this out further with the DWP and HMRC.

122 8.2.6 Interactions with Welfare Benefits A permanent or pilot CBI scheme can interact with social security benefits in a variety of ways. This includes considering which entitlements a CBI should replace and which should remain alongside a CBI. Some advocates for a CBI suggest a complete re-design of the tax and social security system which would remove most existing benefits.71 Others suggest initial modest changes which introduce a CBI payment in addition to current social security benefits.36

Some UK think tanks are critical of models which replace some current benefits with a similar benefit-level CBI rate. Based on economic modelling, the Joseph Rowntree Foundation (JRF) and the Institute for Public Policy Research (IPPR) suggest a CBI scheme akin to the current benefit level risks making relative child poverty worse. JRF suggest a CBI at this level would be of greatest benefit to individuals with modest incomes (i.e. some earnings) and people who are fully dependent on welfare support would not be any better off.72 IPPR point out that this would increase median incomes, thus shifting the relative poverty line and pulling more children into relative poverty.73

While welcoming this insight, the Steering Group is mindful of the limitations of the research. The conclusions by IPPR and JRF were based upon static modelling which did not take account of potential behavioural effects resulting from receipt of a CBI, nor did they include the possibility of aligning a CBI with substantial changes to the tax system. The Steering Group commissioned economic modelling research which enhances a static model by building in anticipated behavioural changes such as labour supply decisions. The Steering Group have used outputs from the commissioned research to inform the feasibility study and contribute to the wider debate about the potential costs and benefits of a national CBI policy (see sections 9 and 10).

The options described below are based on the current welfare system where both Universal Credit and legacy benefits exist and therefore differ from options being tested within the economic modelling research.lviii

As noted previously, a key principle of the CBI feasibility study is to ensure vulnerable and low- income participants are not financially worse-off within a pilot scheme. The total amount which an individual or household receives in benefits is dependent on various factors, including housing costs, income from earnings (if applicable), childcare costs, limited work capability, caring responsibilities, disability or health costs. A CBI pilot which prevented financial detriment to participants would need to ensure these costs were covered, either by including these costs within a CBI payment (potentially making a payment very high) or retaining key benefits alongside a CBI.

8.2.7 Assessment of Scenarios for CBI and Benefit Interaction An overview of the evidence relating to benefit interaction scenarios, associated rationale and the key risks is presented in Appendix 6. These scenarios are based on analysis of evidence from commissioned research, initial engagement with HMRC and DWP and additional research by the Steering Group.

The benefit interaction scenarios which have been considered by the Steering Group are:

lviii It should be noted that for the purposes of the economic modelling commissioned by the Steering Group, the proposed year of analysis assumes that Universal Credit is the main welfare system and legacy benefits will have ended but for a small group of recipients. This was essential to rationalise the number of benefit-interactions options being tested through micro-simulation.

123 Option 1: CBI payment with some entitlements suspended (for both low and high CBI). (Suspended entitlements detailed in Appendix 6).

Option 2: CBI payment with access to all means-tested and non-means tested benefits with CBI income disregarded (for both low and high CBI)

Option 3: CBI payment with access to all means-tested and non-means tested benefits with the current default legislative rules for treatment of CBI payments (for both low and high CBI)

The preferred option of the Steering Group is Option 1: CBI payment with some entitlements suspended. Option 1 reduces conditionality and the impact of means-testing associated with the suspended entitlements. This is considered to be closer to a ‘true’ version of CBI, while minimising detriment and achieving benefit savings, than either Option 2 or 3. The intention of the Steering Group has always been to propose as pure a model of a basic income as possible, in order to test out these principles for the Scottish context. On this basis there are several major institutional challenges associated with delivering a CBI pilot in line with Option 1. These are outlined in detail in Appendix 6 and summarised below:

1. Substantive primary legislation changes required to suspend benefits.

There are significant legislative challenges and risks associated with delivering a CBI which suspends access to a range of entitlements. Ensuring participants are not at risk of detriment by retaining access to some top-ups, premiums and elements will require detailed unpicking of the current social security provision. This will require complex and time-consuming changes to a range of Acts and regulations which are mainly the responsibility of the Department for Work and Pensions and HM Revenue and Customs.

2. There is a risk of financial detriment through loss of access to passported benefits, National Insurance credits and tax reliefs.

Pilot participants may be at risk of financial detriment if passported entitlements cannot be maintained due to suspension of some benefits. This may also result in participants losing access to Jobcentre Plus support services. Losing passporting rights may also include loss of National Insurance credits.

3. Risk of legal challenge associated with different treatment as part of a pilot.

There is a risk of potential legal challenges if legislation was brought in which causes pilot participants to be treated differently to those not involved in a CBI pilot.

4. Clear legislation required to clarify treatment of CBI payments.

This would reduce the risk of inconsistent treatment of CBI for the purposes of calculating benefit entitlement and ensure clarity for decision makers administering different areas of social security.

5. Significant technical and IT barriers.

There are technical IT barriers associated with delivering Option 1 and any other option that required continued interaction. Changes to the current legacy and Universal Credit system would be costly, problematic and involve changes to legislation.

These pose significant challenges for the feasibility of a pilot of CBI as per the model described in Section 5.

124 Option 2 and Option 3 provide context and evidence on alternative benefit interaction scenarios. Both alternative options may be considered more feasible within the current institutional context as they describe scenarios where no entitlements and associated conditionality are suspended, and CBI is paid alongside all existing benefits. The options differ in terms of the treatment of CBI payments for the calculation of means-tested benefits, which brings different risks and issues for a CBI pilot study.

Option 2 (CBI payment with full access to all means-tested and non-means tested benefits with CBI income disregarded) is the surest way to avoid detriment to pilot participants, however it would be of less value in learning about the potential effects of a CBI scheme which may be designed to replace part of the current benefit system. Means-testing and conditional requirements of the current benefits system would remain in place during a pilot of CBI, making it difficult to evidence the effects of a CBI.

Option 3 (CBI payment with access to all means-tested and non-means tested benefits with the current default legislative rules for treatment of CBI payments) would interact with the current benefits system in ways which could have a detrimental impact on participants if CBI was fully counted as income for means-tested benefits. This option would be particularly detrimental to participants whose only income is from benefits as CBI would remove access to passported entitlements, potentially causing a reduction in overall income.lix

At this stage, the Steering Group consider Options 2 and 3 to be of lesser value in exploring the potential effects of a CBI consistent with the principles set out in Section 1 of this report. During a pilot, both options are likely to interact with the current benefit system in ways which retain substantial means-testing and conditionality, particularly in respect of job-seeking requirements. This would erode some of the desired characteristics of a CBI and make the impact of a CBI with those characteristics difficult to evidence.

8.2.8 Interaction with HMRC and Treasury A CBI pilot of the model described in Chapter 5 would also interact with elements of the welfare benefits system for which HMRC have responsibility. These include tax credits, Tax Free Childcare as well as impacts on Child Benefit and potential impacts on High Income Child Benefit Charge. This would require further and substantial engagement with HMRC to explore the issues associated with these interactions.

In order to fund a national CBI policy in Scotland changes to the current tax structure would be necessary. Section 9 provides further detail on what changes may be required as part of economic modelling of a CBI policy. Under current legislation it is not possible to make amendments to taxation for a pilot study of CBI in the same way as would be required for a nationwide policy. Although Scottish Government has the power to set rates and bands of income tax for Scotland, it is not possible to vary these at a local pilot level. There are also legislative and technical barriers lix One example of this, that could clearly be a work disincentive, is the potential that access to childcare costs currently available would be ended. A parent or responsible carer in receipt of universal credit can access up to 85 percent of their childcare costs. This is capped at £646.35 per month for one child and £1108.04 for two-plus children. If the CBI is counted as income it could end the universal credit award and any associated support with childcare costs. Within this option even meeting the childcare costs for one child would not be compensated by a low level CBI (only equivalent to around half the childcare costs) and take up the majority of a CBI paid at the higher level. This is just one of the potential detriments associated with ending passporting within the benefit system that a CBI pilot would have to interact with.

125 associated with amending the tax-free Personal Allowance threshold which is reserved to UK government (currently set at £12,500).

It is therefore necessary to consider different options for modelling and funding a pilot and to determine what is legislatively and technically possible within the present context. As part of a pilot, it would not be possible to reclaim CBI funds from higher earners if the CBI payment was exempt from tax.

A taxable CBI pilot would provide a way of mimicking how a CBI would function if rolled out as a national policy. In doing so, it would allow some of the CBI funds to be clawed back through income tax collection.

There are some key issues to consider while investigating whether a CBI pilot should be taxable. Commissioned research undertaken by CPAG and feedback from HMRC officers identify the following implications of income tax liability:

1. If CBI was considered taxable income, pilot participants would have to notify HMRC of how much CBI they receive and if their income was over the personal income tax allowance threshold. This would likely require completion of a self-assessment tax return, which may be a new process to many individuals. 2. A fully taxable CBI would be fully counted as income for the calculation of means-tested benefits. This would lift some participants out of means-tested benefit entitlement, leading to a loss of passported benefits and risks detriment for pilot participants. 3. A taxable CBI would increase a participant’s liability for income tax and could push some people into paying higher rates of tax, however participants would not be in financial detriment as a result. Liability for income tax would depend on their overall income level (inclusive of a CBI) and whether any taxable benefits were replaced or withdrawn due to a CBI.

Although participants with higher incomes may lose tax reliefs or have new tax charges made due to addition of a taxable CBI payment, they would not be in financial detriment as a result. For example, it could impact eligibility for Marriage Allowance and push higher earners to annual incomes above £50,000 resulting in the tapering of Child Benefit.

The question of taxation is closely linked to the options for delivery of a CBI payment. The delivery mechanisms outlined in Table 16 have varying precedents for the treatment of payments for income tax.

Research by CPAG draws attention to an inter-governmental agreement which suggests if CBI was delivered by Scottish Government via a new benefit (Exception 10), it would not be considered taxable income.lx However, it was also noted that recent legislation was used to clarify that new benefits paid by Social Security Scotland should be non-taxable. This would suggest that although a potential pilot CBI delivered by Scottish Government as a new benefit would likely be considered non-taxable, for the avoidance of doubt, amendments to income tax law would be required to provide clarity.

Similarly, a CBI delivered via local government, charity or private company, would also require legal clarification on whether it was considered taxable for pilot participants. In the case of a charity or lx The agreement between the Scottish government and the United Kingdom government on the Scottish government’s fiscal framework, February 2016, para 89

126 private company, it is noted by CPAG that there may be tax liability for the organisation delivering the payment, which has implications for the attendant administration costs of delivering a CBI.

To ensure clarity and for the avoidance of doubt, it seems likely that in all options for delivery of a CBI, legislation would be required to define whether payments are taxable or not.

8.2.9 Conclusion and Recommendations This section has provided a summary of the key institutional options and challenges associated with delivering a pilot CBI. Under current legislation and delivery powers, there are substantial challenges to delivering a pilot which adequately tests all the principles of a CBI set out in our preferred model, while also ensuring pilot participants (particularly vulnerable and low-income groups) are not in financial detriment as a result of participating in the study.

The key challenges are summarised as follows:

1. Delivery mechanism and power to pay CBI: The current legislative powers do not permit delivery of a CBI pilot of the model proposed by the Steering Group. The available delivery powers would place significant restrictions on the pilot model design. For future discussions it will be important to explore further any legislative issues. 2. Benefit interaction scenario: The Steering Group preferred scenario (Option 1) faces a number of risks and challenges to successful delivery. These include complex legislation changes associated with suspension of benefits, significant technical barriers and risk of financial detriment to pilot participants. 3. Tax interaction: In order to provide relevant and sound evidence on the behavioural and economic effects of a national policy, a CBI pilot would be required to interact with the tax system. A pilot which was not able to interact with the tax system would be of limited use in understanding the behavioural, economic and social effects of a CBI policy. A non-taxable pilot CBI would also not be clawed back from higher earners. However, it will be necessary to clarify through legislation if CBI is to be considered taxable. A taxable CBI could also have a number of implications, including: necessitating participants to notify HMRC of their change of income, a taxable CBI would be counted as income for means-tested benefits (removing some participants’ entitlement to means-tested benefits, risking a loss of passported benefits and financial detriment for pilot participants).

This section has demonstrated that it is not possible to assess tax and benefit interactions as separate entities due to the interdependencies between them. It is essential both are considered together as part of the overall institutional feasibility of any CBI pilot. Any Scottish basic income pilot will need to rely on the full collaboration of DWP and HMRC, with any future engagement taking place concurrently. It is clear from work to date that the challenges of piloting a CBI which has to interact with the existing tax and benefit systems, prohibit at this stage, the testing of all the characteristics in our preferred model.

Notwithstanding these challenges, consideration could be given to whether there are other options that may provide useful evidence about aspects of basic income within the Scottish context. Such other experiments could possibly explore further benefit interaction scenarios 2 and 3 as described in Appendix 6. Our International Learning Report14 emphasised the importance of unconditionality of a CBI pilot due to the potential to have a significant positive impact on participant behaviour and health outcomes. This is supported by research into the effects of current welfare conditionality in the UK15 which recognises the potential negative impacts on health and wellbeing of conditional payments.

127 If alternative benefit interaction scenarios are explored further, it should be noted there will be substantial limitations for learning about the effects of a true CBI in Scotland (appendix 6 outlines these limitations in detail).

In this regard, should any further work be undertaken, it will be necessary to first identify appropriate legislation which is both institutionally feasible and suitable to deliver the specified pilot model. It is recommended that any further engagement with the DWP and HMRC must consider the following: a) Legal and technical assessment of the specified model and delivery vehicle, detailing any legislative, process and IT issues arising. b) Any benefit and tax interactions associated with the model. c) The likely position regarding how a CBI would be treated as income for tax and benefit purposes. d) The options and requirements to facilitate the “switching off” of benefit conditionality for pilot participants. e) Scope out fully any potential detriment issues that may impact on pilot participants and identify potential solutions. f) The issues arising from including those of pension age in a CBI pilot and identify possible solutions.

8.3 Psychological feasibility

This section considers the psychological feasibility of conducting a CBI pilot in communities in Scotland. The analysis rests on the observation, conceptualised by De Wispelaere and Noguera,43 that the legitimacy of any policy implementation, and by extension the success of a policy pilot, requires a broad level of acceptance within the general public for whom the policy applies. The importance of this aspect of feasibility rests on three factors:

• Civic legitimacy: public acceptance and support for the policy are necessary to secure backing and engagement from political groups and civic society agents needed to implement the pilot. As such, psychological feasibility can be considered a key influence on strategic feasibility discussed in Section 8.1. • Democratic legitimacy: the imposition of a CBI pilot in the face of strong opposition by the public and community in the areas selected for piloting could be considered undemocratic and be contrary to the concepts of informed consent and ‘no detriment’ necessary for the pilot to meet ethical standards required. • Practical legitimacy: As De Wispelaere and Noguera43 point out, “illegitimate policies often fail because those subject to them refuse to play by the rules”. In this scenario, the imposition of a CBI pilot in a community within which there was strong opposition would face difficulties in securing participants, maintaining participation throughout the pilot period, and evaluating the experiences of CBI participants.

In order to assess the feasibility of a CBI pilot across these aspects of legitimacy, the Steering Group have assessed the available evidence (outlined in Section 3.4) across three core questions:

128 1. What is the evidence of broad public support and acceptance of CBI, either as a pilot or as a policy? 2. What are the limits to that support, what is the evidence on opposition to CBI, and what constraints to feasibility arise from that opposition? 3. What next steps are required during the feasibility research and during implementation and planning of a pilot to address those constraints and ensure psychological feasibility?

The psychological feasibility of a CBI pilot rests on the test of whether the policy can command sufficient levels of public support and acceptance. The analysis undertaken by the Steering Group found that levels of net approval for CBI in principle recorded by a number of UK and local surveys undertaken in recent years; and that levels of net approval for the concept of a CBI trial in local areas and across the UK where this question has been asked.

The analysis further shows, however, that support for CBI is not necessarily universal nor unconditional in all cases. Support is high among the young, the unemployed and those on low incomes. Net disapproval ratings are recorded for some groups, though, including pension age residents, higher income groups, and the self-employed. Support is highest for models of CBI funded through taxation targeted at high income groups, and lowest when funded through increased general taxation, and though the evidence is not conclusive on this point, some surveys show that support is higher for CBI schemes which are targeted at low income groups.

There are genuine fears amongst some groups, most notably those living with a disability or long- term health condition and carers, that the introduction of a basic income would have potential negative consequences for service provision. These fears are focused on both financial questions over how a CBI would be funded, as well are broader questions of political ideology related to the shrinking of state interventions and a reduction in the provision of core health and social care services. Practical financial questions are raised with concern over funding of a CBI which may give rise to opportunity costs impacting negatively on decisions regarding financing public services and a reduction in universal services. These are very real fears given that a reduction of other social programmes is indeed proposed by some commentators as a possible benefit of introducing basic income.74

These findings are useful and provide an indication of the general psychological feasibility of the CBI pilot in principle. It is important, however, to consider some caveats to that conclusion. Much of the analysis considered in this section is based on UK level survey data, and it is possible that levels of public support may be different in the specific communities selected for a CBI trial. Even where local surveys have taken place, in Fife and North Ayrshire, testing has not been undertaken for the specific model of CBI to be used in the proposed trial – including level of payment, geography, time period and other design factors.

All the analysis above further shows that even in areas where support levels are high (the unemployed and low-income groups), sizeable proportions of the population remain opposed to a CBI in principle.

Taken together, these findings suggest that in order for a successful CBI trial to take place, the following is necessary:

1. In preparation for a launch of a pilot, significant communications activity is undertaken to inform local residents, and address the concerns of those groups which are least in support of the policy. 129 2. In order to deliver that activity, a programme of citizen research is undertaken in the pilot communities as part of the early stages of pilot implementation. This programme should seek to build on the analysis presented here to understand the nuances of local support and opposition for the specific scheme to be trialled and develop messaging and communications tools most appropriate to securing support and approval of local residents.

8.4 Behavioural Feasibility

The available evidence from CBI-type interventions provides important information by which to assess behavioural feasibility – that is, the likelihood of individuals behaving in a way that will lead to the desired outcomes. As already described, there are numerous intended behavioural outcomes of a CBI pilot including: use, and experience, of services; improved opportunities and freedom in life choices (training, education, work, caring); health and wellbeing; labour market participation; and community participation and justice outcomes.

There are also several potential unintended or undesirable behavioural outcomes which form the basis of persistent arguments against basic income. These include: encouraging labour market withdrawal; the nature of CBI as regards promoting state dependence; and reinforcement of gender roles. These potential negative outcomes have been assessed against the available evidence.

This section discusses some of the implications of the available evidence for the behavioural feasibility of a pilot of CBI in Scotland. Potential negative effects or unintended consequences of a CBI on behaviours are also considered. An assessment of all the potential behavioural outcomes of a CBI is important when assessing the overall feasibility of a CBI pilot, as these outcomes may affect the performance or survival of a policy in the longer term.

8.4.1 Implications of evidence on behavioural feasibility Assessing the likely behavioural outcomes is crucial as this will determine the likelihood of the intervention having the impact that is expected and intended. Crucially, the possible unintended or negative outcomes must also be assessed so that the intervention can be designed in a way that avoids or minimises these potential consequences. Clearly, not all intended or unintended consequences that we may expect from a policy may be evident in a short-term pilot. The common behavioural fears related to CBI can be interrogated against the current available evidence and the possible impacts of a CBI in Scotland will now be explored.

8.4.1.1 Withdrawal from labour market

Labour market withdrawal continues to be one of the biggest criticisms of basic income and persists despite evidence largely to the contrary. Evidence on labour market impacts vary from study to study and is difficult to draw from conclusions from the published studies. However, it is important to note that the impacts on all behavioural outcomes, including labour market participation, may differ on roll-out of a long-term basic income policy than is reported in short term experiments or from studies of dividend schemes such as those in the tribal casino studies or Alaska Permanent Dividend Fund.10

It is possible that those receiving a basic income might become more selective when seeking employment, thereby forcing employers to increase wages. Conversely, they could settle for lower wages since they already have a basic income. It is also possible that increased consumption could also lead to greater demand for labour. There is also the potential for people to reduce labour market participation more if payments were permanent, as they would have the security of knowing

130 payments would continue. These potential outcomes may also be affected by other government policies, thus the wider policy context is important in mediating the outcomes of a basic income policy.

8.4.1.2 Nature of CBI

The exact nature of the basic income piloted would have an important influence on behavioural, particularly labour supply, responses. This includes the value of payments, whether they are made unconditionally, and whether they are withdrawn in response to other income. Clearly, if payments were sufficiently high to live on without earned income, it would be financially feasible for some people to withdraw from the labour market. Unconditionality is a potential crucial influence on making decisions around alternative uses of time, such as caring for relatives, starting a business, returning to education, or volunteering.

Any deduction rate as earnings from other sources increase could also be expected to influence decisions around numbers of hours worked. Given the importance of aspects of basic income such as unconditionality and non-withdrawal on behavioural responses, it is highly desirable that any pilot of CBI maintains these crucial fundamental elements.

Finally, the permanence or otherwise of any CBI is also likely to have an impact on outcomes. Evidence from the casino studies suggests that the longer the exposure to the intervention, the greater the potential impacts (in this case, on child mental health, length of time in education).10 Clearly, any pilot would be time limited and even if there was a transition period following on from the pilot, participants would know that is not a permanent CBI and this would likely impact on their behaviours. This is a weakness of many pilot studies and the findings would need to be interpreted with caution in assessing the possible impact of policy implementation.

8.4.1.3 Reinforcing gender roles

The potential impacts of basic income on gender roles are debated and potentially contentious. It may be argued that a basic income might increase financial independence for women thus allowing them to leave unhappy, or even abusive, relationships.75 However, it may also be possible that a basic income could reinforce traditional gender roles by increasing expectations that women should stay in the home and engage in caring and domestic labour. The review level evidence11 does not provide much illumination on this debate. The effects on women’s labour participation vary across studies, and there is little evidence that CBI increases rates of martial dissolution.

The reason for differing impacts on female employment is not clear, but the review authors conclude that labour supply responses for women may be more context dependent than those of men. It is likely that the availability of maternity pay plays a role. Women with young children appear to want to stay at home with their children, and there is fairly good evidence this can lead to improved child outcomes. Lone parents have worse than average health which is attributed to poverty, stress, and role strain.76 This being so, a basic income which provides the opportunity to work less and spend more time in the home with children at key points in time might be considered a positive outcome in such circumstances.

This suggests that the wider policy context and individual circumstances are particularly important when considering issues of gender, especially in relation to labour market participation. The potential impact on the gender pay gap in a Scottish context is unclear from the published evidence.

131 Individual, rather than household, payments may also be reasonably expected to have a gendered impact on outcomes. Individual payments in Alaska were associated with an increase in part-time employment for women, whereas in Iran female employment increased despite household level payments. Thus, the evidence on the impact of basic income on gender issues is mixed and this must be carefully considered in the evaluation of any future pilot.

8.4.2 Behavioural Feasibility Conclusions To our knowledge, CBI has not yet been implemented in any country although there have been several pilots of interventions that meet at least some of the basic criteria for a CBI. There are current tests of different forms of CBI in a few countries worldwide, including Finland, Canada and the Netherlands. Emerging evidence from Finland, and other contemporary pilots, may yield fruitful findings for the Scottish context.

There are some fundamental gaps in the existing evidence base, primarily due to the difficulty of funding, implementing or evaluating a universal, permanent intervention. Most studies have been short-term interventions with dispersed samples of low-income respondents. Whilst there are very pragmatic reasons for the methodological choices made in pilots, there are always limitations on what can be learnt from a pilot approach and the effects of a full, permanent basic income would be different from such pilot interventions. However, it is important that any future pilot in Scotland is designed robustly in order to maximise the chances of generating the expected and desirable outcomes and avoiding unintended or undesirable outcomes.

The outcomes of interest outlined in Section 6.1 along with the available evidence of behavioural responses to a CBI have been carefully considered in the development of the pilot model offered in this report (see Section 5). In other countries where pilots have been carried out, the political context and framing have had considerable impact on the pilot design and outcomes of interest i.e. Finland and Netherlands were primarily interested in supporting people back in to active participation in the labour market thus the pilots have focused on unemployed people and the outcomes of interest are related to employment. Removing or reducing conditionality is considered as having the potential to make significant impacts on participant behaviour and health outcomes.lxi This is fundamentally different from current welfare design and social norms thus it may be difficult for an unconditional approach to be seen as politically and psychologically feasible, although is an important aspect of behavioural feasibility.

The available relevant published evidence suggests that a CBI could impact on a wide range of social, employment and health outcomes, however the current evidence base for CBI is variable and there is a lack of evidence to assess the effects on long-term service use and wider economic impacts. As such, in order to assess more fully the behaviours that would arise from a CBI, it seems that piloting of the policy is merited in Scotland, despite the caveats outlined above.

8.5 Financial Feasibility

A CBI pilot raises several questions regarding financial feasibility. These include the short-term costs of the pilot scheme if a pilot goes ahead and the longer term fiscal and macroeconomic implications if a CBI were rolled out across Scotland as a whole. Trialling an intervention, such as a Citizen’s Basic Income, is likely to involve substantial spending. It is attempting to meet some of the basic living costs for an entire community for the duration of the pilot study. lxi Highlighted by BIEN and our International Learning Report14 (and supported by work looking at Welfare Conditionality15)

132 It is important to note that the direct costs of the payment, and the administration costs of the trial, will be partly offset by savings in terms of benefit payments replaced, and potentially additional tax revenue as incomes are boosted, depending on the rules agreed. However as noted in Section 8.2.9, under current legislation it is not possible to make amendments to taxation for a pilot study of CBI in the same way as would be required for a nationwide policy. Although the Scottish Government has the power to set rates and bands of income tax for Scotland, it is not possible to vary these at a local pilot level. There are also legislative and technical barriers associated with amending the tax-free Personal Allowance threshold (currently set at £12,500). The precise direct costs of the payment will vary, but it must be remembered that, if rolled out, a CBI policy would replace some existing programmes, with the intention of doing so in a fair and more efficient way.

The longer-term implications will not be clear from a pilot because a pilot would not be at a large enough scale nor of long enough duration to generate the full economic impacts of a CBI rolled out across the population as a whole. However, we took the view that to help inform decisions about whether to go ahead with a pilot, some understanding of these potential economic impacts would be useful alongside the results of the feasibility study. Therefore, we commissioned Fraser of Allander Institute (FAI), Manchester Metropolitan University (MMU) and the Institute for Public Policy Research Scotland (IPPR Scotland) to undertake economic modelling work to estimate these effects. This work is summarised in Section 9. The full report was published in June 2020.

8.5.1 Direct CBI Transfer Costs Calculation Below we provide indicative estimates of the direct costs of CBI payments. These are driven by the level of the proposed payment and the proposed sample sizes as outlined in Sections 5 and 6. As such they will be subject to change in response to decisions taken about the proposed sample size, confirmation of the proposed level of payments and any future uplift of benefit rates or the Minimum Income Standard for the pilot years.

The direct payment costs estimated below do not include administrative and evaluation costs. We have not attempted to estimate administrative costs because of the ongoing uncertainty about what form any pilot could take. The limitations identified in the institutional feasibility section on the potential changes to taxes and benefits and the uncertainty over the mechanism by which a CBI would be paid make it impossible to estimate administrative costs at this time. Evaluation costs are considered in Section 6.5.

Table 17 calculates the direct cost of CBI payments per 1,000 people in the sample.

133 Table 17: Estimated Direct Costs of CBI Payments per 1,000 Pilot Recipients

Scottish Population Age Group High CBI Rate Low CBI Rate Distributionlxii 0-15 years 17% £120.48 £84.54 16-19 years 4% £213.59 £84.54 20-24 years 6% £213.59 £57.90 25 years – pension 57% £213.59 £73.10 age Over pension age 16% £195.90 £168.60 Cost per 1,000 £194,931 £89,870 population per week Cost per 1,000 £10.14m £4.67m population per year

Table 18 calculates the cost of a three year pilot based on the recommendations in Section 6 i.e. that we test the impact of both high and low levels of CBI; that we want to be able to examine the scale of impact on our outcomes separately for men and women but not for people with and without disabilities; and that communities smaller than 2,500 are unlikely to demonstrate community-level impacts. The calculations do not include uplifts in years 2 and 3 and they take no account of payments changing as people move up through the age structure or of people entering or leaving the pilot areas due to birth, death or migration.

The table also includes the estimated savings in proportionate terms as a result of the CBI replacing benefits and pensions as specified in Section 5. These estimates are taken from the analyses undertaken to inform the microsimulation in the first stage of the economic modelling study described in Section 9. These are used to calculate the net costs of the CBI scheme proposed.

The direct intervention cost of a pilot based on these assumptions would be £76m over three years for a sample size of 2,500 at the high level of CBI and around £205m for a sample size of 14,600 at the low level of CBI. This would give a total of nearly £281 million over three years for a study including both low and high levels of CBI.

Taking into account cost offsets due to savings on benefits and pensions, net cost of a pilot would be £62m over three years for the high level of CBI and £124.5m for the low level of CBI giving a total net cost in the order of £186m.

lxii 2018 mid-year estimates from National Records Scotland available at: https://www.nrscotland.gov.uk/ statistics-and-data/statistics/statistics-by-theme/population/population-estimates/mid-year-population- estimates/mid-2018

134 Table 18: Estimated Direct Costs of CBI Pilot Study of 3 Years Duration

Direct cost per 1000 Estimated Estimated Cost offset Cost offset Net cost of population per year Sample size Direct Costs due to due to pilots of CBI Pilot benefit pension savings savings High CBI £10.14m 2,500 £76.0m 7% 11% £61.9m Low CBI £4.67m 14,600 £204.7m 15% 24% £124.5m Total £14.81m 17,100 £280.7m £186.4m

The estimated net cost of a CBI pilot of £186.4m over three years clearly depends on the decisions taken about the level of payment and the scale of the pilot study, including any boost to the size of the pilot study population to take account of non-response and attrition. As such it is subject to change as decisions are made on the models proposed in Section 5 and the evaluation issues discussed in Section 6. The aim of these calculations is to highlight the potential scale of the payments required and the way this is driven by decisions that need to be taken in discussion with stakeholders about the aims, the size and the structure of the pilot study.

135 Section 9: Economic Modelling Findings

9.1 Introduction

Piloting a CBI of the kind proposed in this report would have substantial cost implications. Whether these costs are worth incurring depends in part on whether a pilot would generate useful information on the impact of a CBI (see Section 6). It also depends on the likelihood that a CBI of the kind proposed in this report would have a favourable long-term economic impact were it to be rolled out across Scotland. The case for running a costly pilot would be weaker if a CBI is likely to have negative economic consequences.

However, the scale and duration of the proposed CBI pilot would not be sufficient to explore all the economic consequences of a CBI rolled out across the economy as a whole. Therefore, the Steering Group commissioned the Fraser of Allander Institute (FAI), Manchester Metropolitan University (MMU) and the Institute for Public Policy Research Scotland (IPPR Scotland) to model the economic impacts across Scotland arising from introducing the CBI models specified in Section 5.

The aim of the modelling work was to understand better the cost of potential schemes, the impact of the introduction of a CBI on household incomes, the avenues through which it might impact on the economy, either positively (such as reduced poverty and inequality) or negatively (such as increased unemployment and outmigration), and the scale of these impacts. Another outcome of the work is to outline the actions that government (and others) may need to take to minimise the negative impact on the economy from the introduction of a CBI.

9.2 Approach

The work involved three stages combining different analytical techniques:

Stage 1 used a ‘microsimulation model’lxiii to estimate the gross and net fiscal cost of introducing the ‘high’ and ‘low’ levels of CBI specified in Tables 7 and 8 of Section 5, the changes in income tax required to finance a CBI set at these levels and the initial impacts on poverty and income distribution. This provided an estimate of the immediate distributional implications of the CBI and the mechanisms for funding it, before any changes in economic behaviour were considered.

Stage 2 used a ‘dynamic macroeconomic model’lxiv of the Scottish economy to consider how individuals, households and firms might respond to the short term distributional impacts of the CBI estimated in Stage 1. The results from phase 1 were inserted into the macroeconomic model lxiii Microsimulation uses computer models to predict the impacts of changes in a system on the individual units within that system (such as people, households, firms, etc.). The models are based on rules specified in the modelling process showing how different elements within the system are linked to each other. Microsimulation is widely used to evaluate the effects of social and economic policies on the distribution of income. lxiv Macroeconomic models look at the relationships between economic factors across the whole economy such as interest rates, inflation, government spending and unemployment. They can also analyse the relationships between individual behaviours and their impact on the whole economy, such as how changes in wage rates affect employment and prices. Dynamic models allow for the fact that relationships between elements in a model might change as the impacts of a policy or other ‘shock’ to a system work through. They typically involve multiple iterations to estimate how effects of a policy emerge over time as a system adjusts to such a ‘shock’. 136 to estimate how the economy might perform over time, based on a number of assumptions about the way in which key aspects of the Scottish economy – most notably the labour market – might respond to the income and tax changes arising from the CBI. Stage 3 used the results from the long-run macroeconomic modelling in Stage 2 (for example, changes in employment and wages) in a second round of microsimulation, to identify the potential longer term effects of the CBI on the distribution of household incomes and measures of poverty.

A number of assumptions were made in Stages 1 and 3 of the modelling. These are spelt out in detail in the full report from the FAI, MMU and IPPR modelling team.26 They include:

• the retention and suspension of benefits as set out in Section 5 of this report;lxv • abolition of the personal tax allowance such that tax is payable on all income earned above the level of the CBI; • estimation of results for 2023-24 assuming benefits are uprated in line with inflation and assuming Universal Credit is fully rolled out; • full take-up of the CBI; • that the UK-wide poverty line is not affected by the introduction of a CBI in Scotland and changes in poverty are measured against a fixed UK threshold, which is in line with the official poverty statistics in Scotland; • National Insurance Contributions remain in place at the current rate.

The Scottish Child Payment was not included within the microsimulations due to the relatively recent announcement of the policy. The microsimulations also included a non-CBI policy targeted on reducing child poverty.lxvi Under this option, the two-child limit for Universal Credit and the Benefit Cap are abolished, the second and subsequent Universal Credit Child Elements are increased by £40 per week (in 2019-20 prices) and the Child Element for first children is set to equal the new, higher, rate for second and subsequent children. These rates were then uprated in line with prices to 2023-24.

The microsimulations assumed that the revenue required to fund the CBI would be raised from income taxation, the abolition of the personal allowance, and through savings from the benefits and pensions replaced by the CBI. Other forms of taxation could offer the potential to raise the revenue to pay for a CBI, but these were not considered in this study for several reasons:

• other forms of taxation raise a much smaller proportion of the total tax takelxvii and so they would need to increase proportionally more to generate the revenue required to fund a CBI set at the levels proposed in this study;lxviii lxv Note that the modelling used a figure of £163 per week for the low level CBI for recipients of pension age. This is the 2018 rate of pension credit guarantee credit for a single person aged over pension age which was proposed when the modelling was commissioned. This is slightly less than the 2019/20 rate of the New State Pension (£168.60) suggested in the final version of the CBI proposed in the overall feasibility study. lxvi Option 3 in the results section below. lxvii Total VAT revenues in Scotland are estimated at £11.1bn, corporation tax at £4.0bn and council tax at £2.4bn. lxviii It would be possible to use other combinations of tax changes, for example, VAT, National Insurance, Corporation Tax etc. However, whilst the exact effects would be different, the overall structural changes in incentives within firms, households or individuals would be broadly similar. 137 • focussing on a single tax allows for clear identification of the scale of the change required; • some other taxes – such as VAT – are regressive so using them to fund a CBI would undermine one of the main goals of the policy to reduce poverty;lxix • the number of permutations of additional forms of taxation, including new taxes such as carbon taxes, tourist taxes or land value taxes, is very large, such that trying to explore a full range of permutations of different forms of taxation would make the modelling (and any potential tax structure) very complicated and the results difficult to convey; • evidence of the impact of other forms of taxation on the key parameters in the model such as labour supply responses is very limited, in particular the more novel forms of taxation mentioned above.

The microsimulations assume that the CBI is revenue-neutral. Clearly, in practice, governments have the option to run budget deficits to fund public expenditure plans, at least for a time, but this raises a wider set of economic and political debates that are not the focus of this report. It would also mean that there would be an almost limitless number of permutations of funding sources and levels of deficit funding, which would increase the complexity of the modelling process and the results.

The macroeconomic modelling work in Stage 2 made several additional assumptions that are spelt out in the main economic modelling report and again, only the main ones are highlighted here. Crucial assumptions relate to how the labour market responds to changes in the take-home pay of workers. Two different approaches were adopted. In the first, the modelling team explored a range of possible scenarios discussed in the literature, reflecting a range of assumptions about how labour bargaining institutions might react to the changes in the take home pay of the workers they represent. As the microsimulation results below show, there is a mixture of gainers and losers from a CBI once the taxation required to fund the policy is taken into account. Four different wage bargaining scenarios were modelled to reflect how institutions might respond to these gains and losses. Sections 4 and 5 of the macroeconomic modelling report discuss these scenarios and the (limited) evidence on which they are based. Here we summarise the main features of each.

In Scenarios 1 to 3, institutions seek to restore some or all of the take home pay of the workers they represent. This increases pressure on wages, which in turn impacts on the rest of the economy in various ways.lxx The ‘default’ scenario (Scenario 2) is that the wage bargaining positions adopted only take account of the impact of the CBI on workers’ own income. In Scenario 1, institutions ignore the effect of the CBI on workers’ total incomes (including the CBI payment) and only consider the (after-tax) pay differentials between them and other groups of workers, seeking to restore pre-CBI real wages. This leads to greater upward pressure on wages. In Scenario 3, they consider family incomes (including all family members’ CBI payments) as well as workers’ own incomes (CBI and after-tax wages) in the bargaining positions they adopt. This moderates the upward pressure on wages and the negative macroeconomic impacts that flow from that.lxxi lxix Regressive impacts could be reduced if VAT was part of a package of tax measures including more progressive taxes such as income tax but this would reduce the overall progressivity of the package. This is an example of the policy trade-offs highlighted in the discussion section below. lxx The ‘standard assumptions’ in the wage bargaining model are based upon long-term evidence of how individuals have reacted to changes in take-home income in the past. See for example, Layard, R., Nickell, S. Jackman, R., 2005. Unemployment: Macroeconomic Performance and the Labour Market. (Oxford University Press, Oxford.) lxxi Additional variants of the model take account of some groups’ more limited bargaining power. These are included in the main economic modelling report.

138 Scenario 4 is termed a ‘social wage’ model. In this scenario, upward pressure on wages is eliminated because workers fully value the wider social and economic consequences that flow from the CBI, such as narrowed income inequality, reduced poverty and reduced exposure to economic insecurity.

The macroeconomic impacts of these four scenarios were also compared to the option involving changes to Universal Credit described earlier (Option 3 in the microsimulations).

The second approach adopted by the modelling team focused less on how labour market institutions might respond to the changes in wages and taxes due to the CBI, and more on how individuals themselves might respond.lxxii This takes into account so called ‘income’ and ‘substitution’ effects.lxxiii Substitution effects are changes in labour supplied due to changes in the attractiveness of paid work relative to time spent not earning, arising from the levels of remuneration net of taxes. In principle, a CBI could shift this balance in favour of paid work if it resulted in a lower proportion of earnings being taken in tax and/or withdrawn benefits. In this case, a CBI could increase work incentives. For others, the increase in taxation required to pay for the CBI, potentially combined with higher levels of out-of-work income, could reduce the returns to labour and might act as a work disincentive.

Income effects, in contrast, reflect the impact of income itself on people’s labour supply choices (regardless of the trade-off between work and non-work). If people’s choices are based on the total (net) income they receive in relation to their desired level of expenditure, then an increase in income due to a CBI might reduce their incentive to work (or to work more hours).

Substitution and income effects tend to work in opposite directions. For example, a CBI which provides a particular household with a higher level of out-of-work income but also levies a higher rate of income taxation might reduce work incentives as it makes not-working look relatively more attractive compared to working (substitution effect). But to the extent that households care about their overall consumption as well as the trade-off between work and non-paid uses of their time, the income effect of the higher taxation may partially offset the substitution effect in terms of the impact on work incentives. Income and substitution effects would also have different effects on different parts of the income distribution because the changes in marginal effective tax rates and net changes in income following the introduction of the CBI proposed would differ between different parts of the income distribution.

The balance of these effects in practice is imperfectly understood, although as noted in Section 3 of the main report, there is a consensus in the literature that income effects tend to be smaller than substitution effects. However, specific empirical evidence is scarce, in particular from situations where a CBI at the levels proposed in this report has been introduced. Therefore, the modelling work explores a range of assumptions regarding the responsiveness of labour supply to the changes in income estimated in the microsimulation work, looking at substitution effects alone and the joint effect of substitution and income effects.

Different versions of the macroeconomic models estimated the potential effects of the CBI on lxxii The modelling team call the first approach ‘macro-to-macro’. The second approach is called ‘micro- to-macro’ because it explores the way microeconomic choices at the individual level aggregate into macroeconomic effects at the level of the economy as a whole. lxxiii This is the change in how much people choose to work when taxes change (technically the elasticity of labour supply). It involves both the decision on whether or not to work (the ‘extensive margin’) and the decision to work more or less hours (the ‘intensive margin’).

139 migration in and out of the Scottish economy. The modelling work also explored the economic effects of changes in the income distribution arising from CBI and the different consumption patterns of people at different points in the income distribution. In general, consumption amongst lower income groups is more capital-intensive than consumption amongst higher income groups, which is more service-oriented and therefore more labour-intensive. This means that changes in the distribution of income have potential knock on impacts on the rest of the economy (and on the environmental sustainability of different economic impacts, although these are not explored in this work).

9.3 Results

9.3.1 Microsimulation Tables 19 and 20 show the main results of the microsimulation. Table 19 shows the changes to tax rates required under each of the different models of CBI proposed in Section 5, the gross costs of the different CBI models, the sources of savings and revenues from suspended benefits and the net costs of the schemes. Option 1 is the low level of CBI and assumes an equal percentage point increase in the tax rates in each of the five tax bands. The CBI in this option would lead to an increase of eight percentage points in each tax band, with rates ranging from 27% in the lowest income group to 54% in the top income bracket. Option 2 is based on the higher level of CBI payment and assumes an upper limit tax rate of 85%. It was not possible to reach fiscal neutrality with equal increases in taxation in each tax band without breaching the 85% cap. Therefore, Option 2 involves different levels of increase in the tax rates and a range in tax rates from 58% to 85%. Option 3 is the alternative non-CBI policy option targeted at reducing child poverty by abolishing the 2-child limit and increasing the child element of Universal Credit as described above. Tax rates would only need to increase by 6% on the top two tax brackets to fund this change, leading to a range in tax bands of 19% to 52%.

The gross cost of the CBI is approximately £58bn for the high level of CBI and £27bn for the low rate of CBI.lxxiv Each option is fiscally-neutral (i.e. taxes have been raised to fully pay for the CBI after other adjustments to spending). Model 2 balances the budget by increasing the tax rises in the third and fourth tax brackets, which also makes the tax structure even more progressive.

lxxiv To put this in context, the Scottish Government’s annual resource budget – to fund the NHS, police and fire, local government, environmental services, enterprise & skills, universities etc. – was £28.6bn in 2019-20.

140 Table 19: Summary of CBI Models, Changes to Tax Rates and Costs of CBI Net of Savings

Option 1 2 3 Low CBI, flat income High CBI, varied Universal Credit: no tax rate increases income tax rate 2-child limit, child increases element increased by £40 Income tax rate rises +8 percentage points +49 percentage points +6 percentage points on every band on band 3 on the top two bands +44 percentage points on band 4 +39 percentage points on bands 1, 2, and 5 New income tax rates 27:28:29:49:54 58:59:70:85:85 19:20:21:47:52 (%) Gross revenue -£26.7bn -£57.8bn -£1.0bn implications Savings from benefit £4.0bn £4.0bn £0.0bn reductions Savings from pensions £6.3bn £6.6bn £0.0bn Savings from abolition £9.1bn £9.0bn £0.0bn of personal allowance Savings from tax rate £7.2bn £38.3bn £0.9bn rises Net cost -£0.2bn £0.1bn £0.0bn

Benefit savings under options 1 and 2 amount to approximately £4bn. Savings from the replacement of pensions would lie in the region of £6.3-6.6bn and revenue from abolition of the personal tax allowance would be around £9bn. These figures are similar between the low and high payment levels but leave very different funding gaps needing to be filled with increases in income taxation. Model 1 generates an additional £7.2bn from the tax rises that would be required to make the CBI revenue-neutral. Model 2 requires nearly £40bn.

Table 20 shows the impacts of the CBI on poverty and child poverty before taking into account wider macroeconomic impacts. All options have a substantial impact in reducing poverty, model 1 taking 280,000 people out of poverty from a baseline of 1,140,000 (or a 5.4 percentage point fall from a starting point of nearly 22%). Model 2 achieves a much bigger fall of 910,000 people, or a fall of over 17 percentage points from the 22% baseline. Model 3 produces a much smaller reduction of 170,000 people in poverty, a percentage point fall of 3.2%. All models would also lead to substantial reductions in child poverty, bigger in proportionate terms than the reductions in poverty. For Option 3, which focuses on increases in benefits to families with children, the impact on child poverty actually exceeds the impact of Option 1 despite its much lower cost.

141 Table 20: CBI Effects on Poverty and Child Poverty Before Taking Into Account Macroeconomic Effects - Various Scenarios

Option 1 2 3 Change in the number of people in poverty -280k -910k -170k (from a starting point of 1,140k) Change in the number of children in child -90k -250k -100k poverty (from a starting point of 270k) Percentage point change in the number of -5.4 pp -17.3pp -3.2pp people in poverty (from a starting point of 21.6%) Percentage point change in the number of -9pp -25pp -10pp children in child poverty (from a starting point of 28%)lxxv

Figure 6 shows the extent of financial flows under the CBI. Even at the lower payment level they are substantial and affect the bulk of the population in some way. There are over 1.5 million people whose family income would increase by £2.5k per annum or more, and just under 1 million whose families would lose out by the same amount. The gross shift of resources between members of the public would be £26.7 billion per annum, which, as noted earlier, is similar to the Scottish Government’s current annual resource budget.

Figure 6: Number of People Affected by Varying Levels of Change in Annual Family Income

lxxv Figures are given with no decimal place due to the lesser degree of precision possible given the underlying dataset on which they are based.

142 In summary, the microsimulation results demonstrate that the CBI models proposed involve very large financial flows between members of the Scottish population that would achieve substantial reductions in poverty funded by large increases in taxation, in particular at the higher end of the income distribution. Virtually nobody would be unaffected by a CBI, either by net increases in income driven by the CBI itself, or by reductions in take home pay driven by the tax regime required to fund it.

To achieve fiscal neutrality, a low-level CBI would require the abolition of the income tax personal allowance and a tax rate rise of 8 percentage points. It is very hard to achieve fiscal neutrality with a high-level CBI without very high tax rates. Larger families would gain more, but lone parents less than couples with children. The changes would be progressive and make a substantial impact in reducing poverty, although arguably, the effects on poverty in Option 1 are modest in relation to the amount of redistribution taking place. Option 3 shows that there are more targeted ways of reducing poverty, in particular child poverty. The modelled changes in Universal Credit make a similar impact on child poverty as the low rate CBI model without requiring such large shifts of resources, but also without the other potential benefits of a CBI.

9.3.2 Macroeconomic modelling Table 21 shows the macroeconomic effects of the low level of CBI proposed (Option 1 from the microsimulations), without taking into account any migration effects. The results under Scenarios 1 to 4 relate to the different wage bargaining positions described earlier. Scenario 1 assumes that the bargaining position adopted seeks to fully restore real after-tax wage rates generating substantial upward pressure on wages. Scenario 2 adjusts Scenario 1 on the assumption that workers fully value their own personal CBI, recognising that the CBI partially compensates them for the loss of income (including the loss of the personal allowance and some benefits). As a result, they do not seek compensation for these elements in their wage bargaining. Scenario 3 adjusts Scenario 2 assuming that workers’ fully value their own and their family members’ CBI payments in assessing the net change in their income and the wage bargaining positions they adopt. This further moderates their wage demands. Scenario 4 is the Social Wage model, which assumes that workers value all CBI payments, irrespective of recipient, and attach as much weight to others’ CBI receipts as they do to their own disposable income. This curtails wage demands in response to the CBI and mitigates the upward pressure on wages. The final model identifies the macroeconomic impact of Option 3 in the microsimulations i.e. the option involving changes to Universal Credit. Detailed results are not presented for Option 2 from the microsimulations because the macroeconomic models struggled to solve for a net fiscal change of this scale, which exceeds the total of Scottish current government expenditure.lxxvi

The timescale for each of the models varies according to how many years the model predicts it will take for a new equilibrium to arise but is typically in the order of 15-20 years.

The default scenario (Scenario 2) for the low payment version of CBI under the bargaining model with no migration effects leads to a reduction in GDP of 4.4%, relative to what it would have been in the absence of the CBI in the long-run. It is important to recognise that this does not necessarily represent a fall in GDP relative to the current position i.e. it is not necessarily a shrinkage in the economy. Rather, it is a slower growth in GDP than it would otherwise have been in the absence of a CBI. Employment would be 5% less than it would otherwise have been, although this might be spread across the workforce through, for example, a shortening of the working week or lxxvi It is clear, however, that the macroeconomic impacts would be far greater than those reported in Tables 21 and 22.

143 increased uptake of valuable unpaid activities such as volunteering or caring, rather than increased unemployment of those seeking paid work. Consumption in the lowest income quintile would be approximately 29% higher.

Under Scenario 1, workers’ efforts to restore fully the pre-CBI wage rates lead to a bigger negative impact on future GDP (-8.8%) and employment (-9.7%) by increasing the upward pressure on wages in response to the CBI and consumption in the lowest income quintile actually falls slightly compared to Scenario 2. In Scenario 3, workers wage demands are reduced relative to the default scenario (Scenario 2) and as a result, GDP and employment still fall, but by less than Scenario 2. Consumption in the lowest income fifth of the population would be slightly higher than Scenarios 1 and 2.

In Scenario 4 (the Social Wage) there is a small positive effect on future GDP, a small negative effect on employment and an increase in consumption in the lowest income quintile of nearly 31%.

Option 3, involving the changes to the child elements of Universal Credit, leads to falls in GDP and employment of less than 1% compared to what they would have been in the absence of those changes to Universal Credit, but only increases consumption in the lowest income quintile by less than 4%.

Table 21: Summary Macroeconomic Impacts Relative to Future Status Quo of the (Low-Level) CBI Under Various Assumptions About Bargaining, No Migration.

CBI wage bargaining scenarios Option 3 Scenario 1 Scenario 2 Scenario 3 Scenario 4 Percentage Wage Wage Wage Social wage Universal difference bargaining bargaining bargaining model credit relative to no CBI based on adjusted adjusted for changes scenario in the maintaining for workers’ families’ CBI long run (likely to after tax real CBI (default be 15-20 years) wage rates option) GDP -8.8 -4.4 -1.7 0.2 -0.7 Employment -9.7 -5.0 -2.1 -0.1 -0.8 Consumption of 26.7 28.7 30.0 30.8 3.7 lowest quintile

Notes: 1. All the wage bargaining scenarios are based on low level CBI (Option 1 in the microsimulations). 2. Figures are percentage differences between the outcomes under CBI and outcomes under the no CBI scenario. Negative figures for GDP and employment do not represent contractions relative to the current position, they represent how much less GDP and employment would be compared to a no-CBI scenario in the long-run.

Table 22 shows the same set of results with potential migration effects factored in. Migration effects can be both positive, with the possibility that the promise of a CBI attracts people into the economy, and negative, due to the higher rates of taxation on higher earners and the feedback loop as growth and employment fall due to a CBI and act as a deterrent to further in-migration. The

144 modelling suggests that, on balance, these effects are negative, leading to substantially bigger reductions in future GDP relative to the no-migration scenarios, again mitigated by the various factors putting downward pressure on wage bargaining effects. The social wage and Universal Credit models are largely unaffected.

Table 22: Summary Macroeconomic Impacts of the (Low-Level) CBI Under Various Assumptions About Bargaining, With Migration.

CBI wage bargaining scenarios Option 3 Scenario 1 Scenario 2 Scenario 3 Scenario 4 Percentage Wage Wage Wage Social wage Universal difference bargaining bargaining bargaining model credit relative to no CBI based on adjusted for adjusted for changes scenario in the maintaining workers’ CBI families’ CBI long run (likely to after tax real be 15-20 years) wage rates GDP See note -15.2 -5.8 0.1 -1.8 Employment See note -16.4 -6.5 -0.1 -2.0 Consumption of See note 23.7 28.1 30.8 3.4 lowest quintile

Note: Due to the scale of the changes in the migration scenario, the macroeconomic model here struggled to provide a coherent solution. Numbers are not provided here, but the scale of impact would be larger than the other scenarios modelled.

Further models look at the differential effects in different sectors of the labour market defined according to the skill level of the labour force. These are detailed in the main economic modelling report, but they do not alter the broad findings or conclusions reported here. Taking account of the income and substitution effects in the micro-to-macro analyses,lxxvii, lxxviii the macroeconomic impacts are typically significantly smaller than for the macro-to-macro analysis, although they still imply falls in GDP and employment compared to the no-CBI baseline, due in large part to the negative effects on labour supply following the introduction of a CBI, even in the lower part of the income distribution. This is because although CBI policies have the potential to improve work incentives by avoiding the withdrawal of out-of-work and low income benefits as people move into employment, the CBI modelled here retains some means-tested elements of the existing benefit system (including housing benefit and child disability elements of Universal Credit). In addition, the CBI is largely funded through income tax. This means that overall, the CBI and the mechanism proposed to fund it tend not to reduce marginal effective tax rates (or participation tax rateslxxix) compared to the status quo. lxxvii As noted earlier, the ‘macro-to-macro’ analysis looks at the effect of wage bargaining pressures across the economy as a whole. The ‘micro-to-macro’ explores the way microeconomic choices at the individual level aggregate into macroeconomic effects over the whole economy. lxxviii This involves replacing the economy-wide wage bargaining positions assumed in Scenarios 1-4 with the estimates of the individual-level labour supply responses to changes in financial incentives caused by the CBI and the tax changes required to fund it (see Section 3 of the main macroeconomic modelling report). lxxix The participation tax rate measures the proportion of total earnings taken in tax and withdrawn benefits when people move into work.

145 In Stage 3 of the analyses, the results of the macroeconomic modelling were fed back into the microsimulation. Some of the key results from these analyses are presented in Table 23.

They suggest that the effects on poverty and child poverty presented in Table 20 would be reduced.lxxx For example, in the low payment model of CBI (Option 1 in Tables 19 & 20), taking into account income and substitution effects, and the subsequent macroeconomic consequences of these, the impact of the CBI on poverty would fall from a reduction of 280,000 (5.4 percentage points (pp) on a baseline of 21.6%) to a reduction of 250,000 (a reduction of 4.7pp). The effect on child poverty would fall from a reduction of 90,000 (9.0pp) to a reduction of 80,000 (8pp). Both wage bargaining scenarios produce similar, substantial reductions in poverty but also bigger increases in average wage rates and bigger falls in the number of people in paid employment.

Table 23: Effects on Poverty, Child Poverty and Net Cost of CBI After Taking Into Account Macroeconomic Effects - Various Scenarios

Stage 1 Stage 3 Microsimulation results including Microsimulation macroeconomic effects results Option 1 incl. Option 1 Option 1 Option 1 from income and with wage with wage Tables 19 & 20) substitution bargaining bargaining with no macro effects adjusted for adjusted for effects (micro-to- workers’ CBI families’ CBI macro) Effect of CBI on number of NA -90,000 -420,000 -170,000 people in paid work (base = 2,560,000) Effect of CBI on average wages NA +3.7% +24.5% +8.4% Net cost of CBI -£0.2 bn -£0.6 bn -£0.2 bn -£0.2 bn Effect of CBI on poverty (base = -280,000 -250,000 -280,000 -280,000 1,140,000) Effect of CBI on child poverty -90,000 -80,000 -80,000 -90,000 (base = 270,000) Effect of CBI on poverty (base = -5.4 pp -4.7 pp -3.3 pp -4.6 pp 21.6%) Effect of CBI on child poverty -9 pp -8 pp -6 pp -8 pp (base = 27.8%)lxxxi

lxxx Rather than predict what might be the economic situation in the next 15-20 years, given the uncertainties in knowing what the situation would be even in the absence of a CBI, the modelling estimates what the impact would be if the economic outcomes predicted from the three stages of modelling were applied to current levels of poverty and employment. lxxxi Figures are given with no decimal place due to the lesser degree of precision possible given the underlying dataset on which they are based.

146 9.4 Economic Modelling: Discussion

The first point to note in interpreting the results is that modelling does not provide a precise forecast of what the impact of a CBI ‘will’ be. Rather the analytical approach is designed to shed light on the key avenues through which introducing a CBI could impact on the Scottish economy; the underlying costs of different schemes; the likely scale of different effects; the potential positive and negative implications of different scenarios for how people and businesses might respond; and most importantly, the types of behaviours that could drive particular outcomes. As such, the single point-estimates of specific impacts are less useful than the range of outcomes, highlighting the possible scale and direction of effects under various assumptions and scenarios, and elucidating the drivers of the modelled effects. This is true in macroeconomic modelling in general. It is especially important in this case given the scale of the changes that have been modelled and the limited experience there is anywhere in the world of such a major structural change to the economy. Less radical CBI options could have been considered but Section 5 of this report provides the rationale for the options the modelling team were asked to consider.lxxxii How individuals, households, businesses and the entire economy respond to such radical changes – not just in their own circumstances but across society as a whole – necessarily made the modelling more uncertain than it would be when modelling the smaller scale changes we usually see in policymaking.lxxxiii

The second key point is that the scale of the effects depends on how society, including government, responds to such a fundamental change in the economy. All models lead to substantial reductions in poverty and inequality which would need to be funded by increases in taxation. Most models lead to slower growth and reductions in overall employment compared to a situation with no CBI. However, the scale of these effects depends on how individuals and institutions respond to the CBI, to the associated tax changes and to the effects on their absolute and relative income. There are a number of uncertainties here. Firstly, some people may not respond – or have the opportunity to respond – to a substantial increase in the amount of tax they pay in the manner suggested in the modelling work. Scenario 4, the Social Wage model, shows the potential impact if labour supply responses were very muted, but the evidence on such an outcome is particularly limited, and evidence on labour supply responses in general, particularly over the long-run, is imperfect.77 Secondly, as noted above, the introduction of a CBI may have positive impacts upon labour supply incentives. It will, for example, remove the high effective marginal tax rates that some people face in returning to work, particularly those on relatively low incomes. It won’t remove them completely, because recipients of housing, disability and childcare benefits, which would be retained under the CBI models proposed, will continue to face high effective marginal tax rates as these would continue to be means-tested and tapered.lxxxiv Nevertheless, for some it will reduce the disincentives to work that come from very higher effective marginal tax rates. Finally, and most importantly, if society’s preferences are such that people are willing to pay higher taxes in exchange for a significant reduction in inequality across society and the other potential benefits of a CBI, or to put in another way if people support the idea of a ‘social contract’, then the labour supply response may not be as stark as many of the modelled scenarios suggest. lxxxii In practice, less radical options could be considered as part of a gradual introduction of a CBI to manage some of the effects outlined below, but they would not test a CBI true to the principles defined earlier in the report. lxxxiii Such as, for example, the changes in the tax bands and tax rates implemented by the Scottish Government since the devolution of income tax in 2016. lxxxiv Whilst this is an impact for a large number of people in the economy, from a macroeconomic perspective the cumulative impact – whilst important – is unlikely to be the most important factor driving the results. Instead it will be the impacts on the population as a whole. 147 The third key point, related to the second, is that if workers are enabled by the CBI to enhance their skills by taking time out of work to undergo training, the CBI may boost productivity. Improved health and other social outcomes for workers may enhance this effect by reducing absenteeism. These boosts to productivity could offset some of the effect of increased costs and reduced competitiveness. The impacts of increases in productivity are explored in Section 4.6.4 in the main macroeconomic modelling report. The analyses suggest that to offset the GDP contraction in the default scenario, Scenario 2, in which workers fully value their personal CBI, productivity would have to increase by 4.3%. This is a substantial shift and government would need to consider how this could be achieved using supportive policies in parallel to a CBI to boost labour productivity. It is particularly challenging given that some of the productivity effects of a CBI are potentially negative due, for example, to the disincentives it might create for many workers (see Section 3 of the main modelling report). Unfortunately, the empirical evidence on the size of potential productivity effects of a CBI remains limited.

The final key point in interpreting the results of the modelling work is that there is no right or wrong answer in the macroeconomic arguments for or against a CBI. The policy involves trade- offs between potentially competing goals. At the risk of oversimplifying the trade-offs involved, the modelling suggests that the CBI proposed would reduce poverty, child poverty and income inequality. It would also have the potential to reduce economic precarity and change the nature of the labour/leisure/training/creating/caring choices that people currently face. Slower growth and redistribution of incomes may reduce the carbon footprint of the economy (although as noted earlier, the net carbon effects of the change in income distribution that might arise from the CBI models proposed are ambiguous). However, the modelling also suggests that these potential benefits could come at a cost in terms of economic growth and reduced real incomes for the richest groups. Although the modelling does not necessarily suggest a shrinkage of the economy compared to the current position, most scenarios suggest that the economy would be smaller than it would have been in the absence of a CBI. This has a number of implications, for example, for Scotland’s relative position compared to neighbouring economies, some of which are closely integrated economically with the Scottish economy. This might, in turn, affect migration and relative competitiveness, and increase the potential negative impacts of a CBI. An economy growing more slowly has other important implications, for example, regarding the ability to generate more resources to support an ageing population.

The modelling cannot eliminate these uncertainties entirely nor indicate which of these value judgements should guide policy, but it does highlight the impact of different assumptions about how society and the economy might respond to a CBI and the trade-offs between policy goals that need to be considered. It suggests strongly that, given the scale of the tax increases (or other funding sources) required, a high level of support for a CBI (and a commitment to pay for it) would be required across all sections of society to maximise the potential benefits of a CBI and minimise the potential risks.

148 Section 10: Feasibility Study Conclusions and Recommendations

This project set out to design a CBI pilot to test the role of a CBI in reducing poverty in Scotland and explore the feasibility of local basic income pilots. The Steering Group did this by gathering and synthesising CBI evidence across published research, learning from contemporary pilots, community engagement and survey data, and commissioning new research to address evidence gaps. This provided the Steering Group with a robust evidence base on which to develop the proposed pilot model, alongside an evaluability assessment to assess the best way of measuring whether a CBI would deliver on the outcomes of interest. This final section summarises the findings of the feasibility study, including modelling of the potential impacts on the wider economy in the longer term, and draws out implications to inform future discussions about CBI and social security more broadly in Scotland.

10.1 Pilot model rationale and recommendation

Potential CBI models differ in terms of the size and characteristics of the eligible population(s), the amount payable, the total cost and potential outcomes, all of which have implications for the evaluation design. Our preferred pilot model attempts to adhere to the overall principles of CBI whilst functioning alongside some existing social security arrangements to minimise the risks of financial detriment to vulnerable and low-income groups. A summary of the key design features of the preferred CBI pilot model is included below in Table 24. A more detailed version is provided in Appendix 2.

149 Table 24: Overview of Preferred CBI Pilot Model

Design Feature Preferred CBI Pilot Model Duration of pilot Paid for three years in the pilot after a one year preparatory period. Experiment Type Saturation within specified area/s. All residents within a geographical area will be eligible to participate in the study. Payment type and regularity • CBI paid by bank transfer or equivalent. • Regular payment (preference for weekly, fortnightly or monthly options). • Given prospectively. Level of payment High level CBI (per week): 0 to 15 years (payment to main carer / parent) £120.48 £120.48 16 years to pension age £213.59 Pension age £195.90 Low level CBI (per week):lxxxv 0 to 15 years (payment to main carer / parent) £84.54 16-19 years £84.54 20-24 years £57.90 25y to pension age £73.10 Pension age £168.60

lxxxv Low level CBI rates reflect existing benefit entitlements: 0 – 15 y = £84.54 (Rate of child tax credit family rate & 1st child rate (£63.84) plus Child Benefit eldest child rate (£20.70)). 16-19 y = £84.54 (Reflecting rate of 16-19 year-old who is still in approved education: Rate of child tax credit family rate & 1st child rate (£63.84) plus Child Benefit eldest child rate (£20.70)). 20-24 y = £57.90 (Rate of jobseeker’s personal allowance for a single person aged 16-24). 25y to pension age = £73.10 (Rate of jobseeker’s personal allowance for a single person aged 25 or over). Pension age = £168.60 (Rate of the new state pension).

150 Design Feature Preferred CBI Pilot Model Interaction with tax system CBI would be included in the calculation of income for tax purposes. CBI would only be taxed if a participants’ total taxable income exceeded the Personal Income Tax Allowance threshold for the pilot year(s). Interaction with other benefits For the duration of the study, pilot participants who would normally be in receipt of elements paid within Universal Credit, and premiums and additions within Pension Credit and legacy benefits should be able to claim these alongside a CBI: Specifically, those relating to disability, limited capability for work, housing, childcare and caring. For pilot participants in receipt of means-tested benefits, CBI should befully disregarded as income for means- tested benefit calculation. A preferred model would be to suspend participant access to the following benefit entitlementslxxxvi for the duration of the study: • Income Support (Personal Allowance) • Income-based Jobseekers Allowance (Personal Allowance) • Income-related Employment and Support Allowance (Personal allowance) • Child Tax Credit (Family Element plus Child Element) • State Pension • Child Benefit • Carer’s Allowance (Basic Rate and Scottish Supplement) • Universal Credit: Standard allowance for Single person • Universal Credit: First child / subsequent child payments

lxxxvi Suggested suspensions are set out in the context of the CBI characteristic of payments being made to the individual. It is not the intention that any couple rates would continue to be paid within these existing benefits. Instead, within this model, a couple would each receive the CBI payment.

151 Design Feature Preferred CBI Pilot Model Number of Intervention Sites / Number of intervention sites Sample Sizes • We recommend testing two payment levels – a low and a high CBI. This would require two intervention sites, one for low CBI and one for high CBI. • We recommend piloting each level of CBI (low and high) in communities typical of Scotland in terms of the primary outcomes. Intervention site sample sizes • An intervention site for the high level of CBI would require a sample size of at least 2,500 people. • An intervention site for the lower level of CBI would require a sample size of 14,600 people.

152 10.2 Evaluation & research considerations

10.2.1 Evaluation Recommendations An evaluability assessment is a way of working through whether and how a policy or an intervention such as a CBI can be effectively evaluated. It involves clarifying with stakeholders the intended and unintended outcomes of the policy and assessing whether and how these can be measured with the time and resources available. The evaluability assessment process has explored the nature of possible pilot models, their likely costs, the potential outcomes, and the hypotheses or research questions that a pilot would seek to address.

Full details of the evaluation recommendations can be found in Section 6.6. A summary of the key recommendations is provided below: a) Any pilot should be a randomised controlled outcome study with two saturation intervention arms (one receiving the high payment, the other receiving the low payment) as this offers the best way of understanding the potential impact of a CBI on a range of social and economic outcomes. Within the pilot communities, everyone receiving the CBI would be invited to take part in the evaluation. b) Based on the theory of change, the primary outcomes should be changes in poverty, child poverty and unemployment. Secondary outcomes should include community level social and economic effects, improved health and well-being and improved experience of the social security system, with the final list developed over time as the theory of change evolves according to emerging evidence and/or new areas of interest. c) The study should be delivered alongside a control area, comprising a stratified random sample of the population drawn from the same sampling frame as the pilot communities. It is recommended the evaluation is complemented by a process evaluation exploring the mechanisms by which any change in outcomes came about, including any facilitators and barriers to the successful implementation of the CBI and any differential effects between different groups of recipients. d) People leaving and people entering the study area should be included in the pilot, with consideration given to various eligibility criteria to reduce the risk that differences in the availability of the CBI between study and neighbouring areas affect the outcomes by distorting decisions regarding uptake of employment, moving area etc. Children born in the study area during the study period should also be included in the pilot and receive the CBI.

The recommendations above specify what we believe to be the most robust evaluation possible. However, there are substantial challenges and important limitations of a pilot and outcome evaluation on the scale and using the design proposed in this report. Limitations of the proposed pilot and evaluation include: a) the limits on generalisability to other areas b) the limited scope to explore the role of contextual effects due to the limited range of contexts (e.g. geographical, social, economic) in which the CBI pilot would be implemented c) the risk that single sites might be contaminated by economic shocks peculiar to those areas

153 d) some of the potential impacts of a CBI are unlikely to occur because of the time-limited, geographically-narrow focus of the proposed pilot and because the institutional barriers might prevent all the features of the CBI proposed being tested.

These are important limitations of a pilot accompanied by an outcome evaluation on the scale and using the design proposed in this report, which need to be considered against the cost of the pilot and evaluation and the evaluation. However, they also need to be considered alongside the strengths and weaknesses of the alternatives, for example, a smaller scale pilot and/or a modelling approach, both of which have limitations of their own.

10.2.2 Policy Governance and Research Ethics A range of ethical considerations have been identified in relation to both the piloting of a CBI in a Scottish context, and the evaluation of such a pilot. Implementing and evaluating a policy pilot based on giving some people or communities a CBI (the intervention group) and comparing the outcomes against a control group who do not receive a CBI has implications for policy governance and research ethics. We would consider any pilot that was implemented or evaluated in ways that were unethical would be infeasible.

Ethical issues relating to piloting a CBI may include: consideration of mandatory participation; conditions required for ‘no detriment’; implications for participants during and post-intervention; issues of potentially withholding a beneficial interventions from controls; ensuring informed consent for both intervention and evaluation; incentivising the intervention and control groups to participate in the evaluation; and implications of linking data from statutory organisations. Policy governance is about the need to protect the intervention group(s) from harms they would not be exposed to if they were not part of a pilot. This includes assessing and mitigating potential negative differential impacts based on a protected characteristic or socio-demographic status.

There are a number of recommendations associated with ethical considerations of progressing a pilot and evaluation as described in earlier sections. A summary of the key recommendations is provided below: a) Participants, particularly those who are vulnerable and/or on low incomes, should not experience detriment (financial or otherwise) compared to individuals not involved in the study. b) Early consideration is given in the development and design of any CBI delivery mechanisms, including the processes for identifying CBI recipients where the delivery organisation is responsible for more than just CBI, and for obtaining consent for participation in the evaluation. c) A transition strategy to support all pilot participants before, during and post pilot should be developed to manage the risks associated with transitioning on or off a pilot or following changes in participant circumstances. d) Prior to the commencement of the study and evaluation, the following steps should be taken: i. Appropriate legal and procedural advice sought to ensure that the selection of intervention areas (and exclusion of other areas), and the mandating of people to participate in a pilot is within the legal competence of the policymaking body, procedurally fair and reasonable; ii. Ethical approval for the evaluation is obtained from an appropriate Research Ethics Committee;

154 iii. The agreed pilot model is assessed in full using the comprehensive Integrated Impact Assessment process; iv. Scottish Government undertakes a Data Protection Impact Assessment; v. For data linkage, data sharing agreements are put in place between relevant agencies at an early stage and a submission is made to the Public Benefit and Privacy Panel.

10.3 Discussion of overall feasibility

There are several, interdependent aspects of feasibility which the Steering Group have used to test the viability and achievability of a pilot. A CBI pilot must be politically and financially feasible to allow successful testing. Political feasibility is an overarching term to describe whether there is appetite for a particular policy, however to make sense of the wide range of political influences, it can be broken down into: Strategic, institutional, psychological and behavioural feasibility.43

Strategic feasibility is the action required to build a robust political coalition of support for enabling the legislation and subsequent institutional arrangements (including financial pathways) to implement a CBI pilot. In recent years, CBI has gathered interest from various UK political parties, albeit at varying degrees. The Scottish Government announced in the 2017 Programme for Government that it would support local authority areas to explore the feasibility of a CBI scheme, a commitment which led to the Steering Group undertaking this work. Basic income features in recent Green Party and Labour Party manifestos, and prior to the 2019 UK General Election, a number of Liberal Democrat candidates signed an open letter in support of trialling an unconditional minimum income element in the social security system.lxxxvii The Steering Group have engaged with UK civil servants to explore the feasibility of piloting basic income in Scotland but, currently, basic income policy forms no part of the UK government programme.

Analysis by the Steering Group has concluded that while varying degrees of interest in CBI exists across the UK political spectrum, there is little evidence of coherence around a view of the correct model and features of a basic income, or its relationship with the existing welfare state. Piloting basic income will require sustained political support across all levels of government (local, Scottish and UK) throughout the duration of the pilot and evaluation. This is essential to overcome the substantial legislative barriers required to make a pilot study institutionally feasible.

Institutional feasibility is important for both practical implementation issues and political institutional support from a range of organisations. Any Scottish CBI pilot will require full collaboration of the DWP and HMRC. The Steering Group have engaged with the DWP and HMRC to explore institutional opportunities and challenges. Under current legislation and delivery powers, there are substantial challenges to delivering a pilot which adequately tests all the principles of a CBI set out in our preferred model, while also ensuring pilot participants (particularly vulnerable and low- income groups) are not in financial detriment as a result of participating in the study. Reducing the scale or scope of a CBI pilot, or amending the model design principles would not eliminate these challenges and would likely result in substantial limitations for learning about the effects of a true CBI in Scotland.

Addressing these challenges would require substantive primary legislation and regulation changes which are mainly the responsibility of the DWP and HMRC. Although complex and time-consuming, changes to legislation are largely dependent on strategic political will and interest. As a result, lxxxvii https://thoughtsofprogress.wordpress.com/2019/11/10/lib-dems-backing-minimum-income-pilots/

155 strategic political engagement alongside further detailed discussion with both the DWP and HMRC will be essential to consider how key barriers to institutional feasibility could be overcome.

Psychological feasibility encompasses the public support and acceptance for CBI, both in terms of whether it is readily understood and seen to be beneficial by communities. Public support is essential to secure strategic backing and engagement from political groups and as such is closely linked to strategic feasibility.

The analysis undertaken by the Steering Group found that although recent UK and local surveys demonstrate net approval for the principle of CBI, support for CBI varies according to different population groups. For example, there is high support among young citizens, people who are unemployed, or on low incomes. There is net disapproval for some groups, including pension age residents, higher income groups and people who are self-employed. Public support also varies according to the method of funding a CBI, with greatest support for CBI models funded through general taxation targeted at high income groups. Although analysis provides a general indication of psychological feasibility for a CBI, it is important to note that public opinion on the model specified by the Steering Group has not taken place and support may therefore differ according to design features such as level of payment, communities included in the study, duration of a pilot and interactions with tax and benefit systems.

Variation in public support according to population group or design feature is broadly reflective of the distinctions in support and engagement from different UK political parties. As such, it demonstrates the need to focus and unify both public and political support around a preferred model in order to strengthen psychological and strategic feasibility.

Behavioural feasibility concerns the likelihood of individuals involved in a CBI intervention behaving in a way that will lead to the desired outcomes. Behaviour changes could either be positive, negative, intended or unintended and may affect the performance or survival of a policy in the longer term. As a result, it is closely linked to psychological feasibility because expectations of behavioural impacts (such as changing labour market behaviour) may undermine public opinion of the policy, which could in turn affect strategic feasibility.

The ways in which people respond to a CBI are largely in response to the model design and institutional arrangements. In countries with CBI-like pilots, the political context and framing have had considerable impact on the pilot design and outcomes of interest. For example, Finland and the Netherlands were primarily interested in supporting people back into active employment thus the pilots have focused on unemployed people and the outcomes of interest related to employment. Evidence suggests that removing or reducing conditionality is considered as having the potential to make significant impacts on participant behaviour and health outcomes.lxxxviii

The available relevant published evidence on behavioural effects suggests that a CBI could impact on a wide range of social, employment and health outcomes. However, the current evidence base for CBI is mixed and there is a lack of evidence to assess the effects on long-term service use and wider economic impacts. In order to assess more fully the behaviours that would arise from a CBI in Scotland, we consider that piloting of the policy is merited.

As noted at the outset of this section, there are multiple interdependent aspects of political feasibility which the Steering Group have used to assess whether a CBI pilot is feasible within lxxxviii Highlighted by BIEN and the International Learning Report case studies14 (and supported by work looking at Welfare Conditionality15)

156 the current Scottish context. The four aspects are interdependent in the sense that there must be strong evidence of feasibility in all cases to ensure there is sufficient political (and public) will and confidence for a successful and robust pilot to take place.

Political feasibility is also strongly related to financial feasibility, particularly in relation to institutional arrangements for tax and benefit systems. The net costs of a CBI pilot are dependent upon the level of CBI payment(s), number of people involved in a trial, experiment administration and evaluation costs, savings due to replacement of benefits and pensions, and the suitability of funding sources. Depending on the rules agreed, additional tax revenues may be boosted and contribute to pilot funding. However as noted in Section 8.2.9, under current legislation it is not possible to make amendments to taxation for a pilot study of CBI in the same way as would be required for a nationwide policy. Although the Scottish Government has the power to set rates and bands of income tax for Scotland, it is not possible to vary these at a local pilot level. There are also legislative and technical barriers associated with amending the tax-free Personal Allowance threshold (currently set at £12,500).

Taking into account the cost offsets due to savings on benefits and pensions, it is estimated that the net cost of a pilot of CBI as described in Chapter 5 (where both levels of CBI are tested), would be around £186m over three years. This is based on approximately 17,100 people receiving a CBI (2,500 receiving high CBI and 14,600 receiving low CBI) and does not include administration and evaluation costs, nor the costs of boosting the sample to account for potential non-response and attrition.

Overall there are particularly strong dependencies between institutional, strategic and financial feasibility. In the Scottish and UK context, institutional feasibility is sensitive to strategic decisions over time as political strategy shapes the development of delivery institutions which are essential for a Scottish pilot of CBI. At this time, it is not currently feasible to progress to a pilot model of CBI as described by the Steering Group due to the substantive and complex legislative and delivery challenges. To strengthen overall feasibility for a CBI pilot, it will be essential to ensure there is strong political engagement at all levels of government (local, Scottish and UK) and across institutional agencies to consider how key institutional challenges could be overcome.

10.4 Economic Modelling Findings

We acknowledge that a CBI pilot of the kind proposed in this report would have substantial cost implications. We also note that the scale and duration of the proposed pilot would not be sufficient to explore all the economic consequences of a CBI if it were rolled out across Scotland. Therefore, the Steering Group commissioned the Fraser of Allander Institute (FAI), Manchester Metropolitan University (MMU) and the Institute for Public Policy Research Scotland (IPPR Scotland) to model the economic impacts across Scotland arising from introducing the CBI models specified in Section 5.

The aim of the modelling work was to understand better the cost of potential schemes, the impact of the introduction of a CBI on household incomes, the avenues through which it might impact on the economy, either positively (such as reduced poverty and inequality) or negatively (such as increased unemployment and outmigration), and the scale of these impacts. The research team also explored actions that government (and others) may need to take to minimise the negative impact on the economy from the introduction of a CBI.

The economic modelling was done over three stages:

157 Stage 1 estimated the immediate distributional implications of the CBI and the mechanisms for funding it, before any changes in economic behaviour were considered. In summary, the stage 1 results demonstrate that the CBI models proposed involve very large financial flows between members of the Scottish population that would achieve substantial reductions in poverty funded by large increases in taxation, in particular at the higher end of the income distribution. Virtually nobody would be unaffected by a CBI, either by net increases in income driven by the CBI itself, or by reductions in take home pay driven by the tax regime required to fund it.

Stage 2 considered how individuals, households and firms might respond to the short term distributional impacts of the CBI estimated in Stage 1 and used these to estimate how the economy might perform over time. This was based on a number of assumptions about how key aspects of the economy, including the labour market, might respond to the income and tax changes arising from the CBI. Different sets of assumptions gave arise to different scenarios, but the central default scenario was of a low payment version of CBI with the assumption that workers fully value their own personal CBI, recognising that the CBI partially compensates them for the loss of income (including the loss of the personal allowance and some benefits) and as a result, they do not seek compensation for these elements in their wage bargaining. Under this scenario, the modelling suggests a reduction in GDP of 4.4%, relative to what it would have been in the absence of the CBI in the long-run. Employment would be 5% less than it would otherwise have been, although this might be spread across the workforce through, for example, a shortening of the working week or increased uptake of valuable unpaid activities such as volunteering or caring, rather than increased unemployment of those seeking paid work.

Stage 3 used the results from the long-run macroeconomic modelling in Stage 2 (for example, changes in employment and wages) in a second round of microsimulation, to identify the potential longer term effects of the CBI on the distribution of household incomes and measures of poverty. Key results from these analyses suggest that the effects on poverty and child poverty presented in stage 1 would be reduced over the longer term.

For example, in the low payment model of CBI, taking into account income and substitution effects, and the subsequent macroeconomic consequences of these, the impact of the CBI on poverty would fall from a reduction of 280,000 (5.4 percentage points (pp) on a baseline of 21.6%) to a reduction of 250,000 (a reduction of 4.7pp).

These findings are important as they suggest that a CBI has the potential to have substantial impacts on the economy in the longer term. The modelling suggests that the CBI proposed would reduce poverty, child poverty and income inequality. It would also have the potential to reduce economic precarity and change the nature of labour / leisure / training / creating / caring choices that people currently face. However, the modelling also suggests that these potential benefits could come at a cost in terms of economic growth and reduced real incomes for the richest groups. Although the modelling does not necessarily suggest a shrinkage of the economy compared to the current position, most scenarios suggest that the economy would be smaller than it would have been in the absence of a CBI.

It is necessary to remember that modelling does not provide a precise forecast of what the impact of a CBI will be, rather it is designed to illuminate some of the pathways by which introducing a CBI could impact on the Scottish economy. A CBI represents a major structural change to the economy and as such single point estimates of specific impacts are less useful than highlighting the possible scale and direction of effects across a range of outcomes. Another key point is that

158 the scale of effects depends on how both society and government policy respond to such a fundamental change in the economy. It is also possible that productivity may actually be boosted in the longer term if a CBI enables workers to enhance their skills by taking time out to undergo training.lxxxix These boosts to productivity could offset some of the effect of increased costs and reduced competitiveness. Finally, the economic modelling work overall highlights that there is no right or wrong answer in terms of the economic impact of a CBI, rather the policy involves trade-offs between potentially competing policy goals.

Overall the modelling suggests that a CBI would result in a slowing of the economy compared to growth in the absence of a CBI, and a reduced real income for the highest income groups. However, the modelling also suggests that the CBI model proposed would reduce poverty, child poverty and income inequality. It would also have the potential to reduce economic precarity and the choices that people face in regard to working, caring, training and creating.

10.5 Learning and discussion

This final section outlines the learning from the feasibility study, including lessons on the approach taken, potential areas for further exploration, and a discussion on where the learning from this work fits in to broader discussions on the future of the Scottish economy and society.

10.5.1 Feasibility study learning The proposed pilot model was designed alongside a comprehensive evaluability assessment leading to the development of an outline theory of change for CBI in Scotland, and an outline evaluation for a future pilot. To address important evidence gaps, the group commissioned two new pieces of research: Exploring the Social Security Implications of a Citizens’ Basic Income Pilot by the Child Poverty Action Group in Scotland; and, Modelling the Economic Impacts of a Citizens’ Basic Income in Scotland by the Fraser of Allander Institute (with Manchester Metropolitan University and IPPR).

Throughout the feasibility study, we remained of the view that due to the current gaps in knowledge and the limitations of pilots carried out elsewhere, there would be little added value in conducting a pilot which tested limited aspects of CBI. As such, we have proposed an aspirational and robust research plan, which would provide an opportunity to pilot and evaluate all aspects of CBI within the Scottish context.

In progressing this work, the feasibility study identified several substantial obstacles that would need to be addressed to enable a pilot testing a universal, unconditional CBI to go ahead. Consequently, we have concluded that in the short to medium term, it is not currently feasible to pilot the proposed model of CBI in Scotland. As noted in Section 10.3, in order to address the key institutional challenges and resultant complex legislative and delivery barriers, there requires strong political engagement at all levels of government (local, Scottish and UK) and across institutional agencies.

lxxxix A collection of essays published by Carnegie UK Trust suggests that better opportunities for training and skills development may contribute to improved work productivity. Report available at: https:// d1ssu070pg2v9i.cloudfront.net/pex/carnegie_uk_trust/2020/01/13104243/Can-good-work-solve-the- productivity-puzzle-FINAL.pdf

159 10.5.2 Next steps Whilst this feasibility study may not result in a pilot of CBI in Scotland in the short to medium term, there has been a great deal of learning from both the results, and the process of undertaking the feasibility study. Throughout the study several opportunities to build on the learning or adopt similar approaches were identified: f) Feasibility study approach – detailed exploration of the feasibility of a policy (or in this case, a pilot) has been useful in identifying in advance a wide range of issues in relation to implementation, testing outcomes, ethical treatment of participants or those affected by the implementation of a policy pilot, and costs. While it is usual to consider the design, evaluation, costs and potential outcomes of an intervention ahead of implementation, this study has shown the enormous benefit of considering broader aspects of feasibility in a systematic manner. The likelihood of an intervention becoming public policy is dependent on a wide range of influences and conditions, covering a range of complex and interdependent aspects of feasibility. For a policy to be successful, the outcomes must meet the aims and objectives set out in the policy intention, and achievable in the given context.41,42 The novel, systematic approach used here to assess aspects of political, ethical and financial feasibility alongside design and evaluation has been constructive and informative. This approach could be adapted and used to systematically explore other policies or interventions in advance of implementation as part of due diligence. g) Exploring knowledge gaps and findings from economic modelling – the novel multi-level approach taken in the CBI economic modelling programme (described in Section 9) sets out new ways to approach future economic modelling, with modelling at the macro level taking account of micro-modelling scenarios. Further work to refine the assumptions underpinning this modelling could be undertaken, as well as adapting the approach for modelling other interventions and economic scenarios that might meet the goals of a CBI without some of the challenges and costs of the CBI proposed in this work. There is also possible merit in exploring policy responses to mitigate or offset the economic shock introduced by policies such as CBI. h) Providing detail on how of a CBI may interact with the social security system – the assumption that it is feasible to pay a CBI alongside other welfare support was explored in detail and in the context of the UK social security system. It would be essential to the feasibility of any CBI pilot that more work is carried out by Scottish and UK governments on benefit interactions that details realistic options to avoid detriment and allow important entitlements to continue. This is not just relevant to a pilot as it is widely argued that even a full CBI would need to interact with other sources of welfare support to meet the additional costs associated with housing, childcare and disability. i) Further public engagement work – it is likely that any future exploration of CBI in Scotland would need to involve further engagement with the public, and specifically any local communities affected by a pilot CBI. Formal consultation may be required, and it is advisable that a formal Integrated Impact Assessment be carried out ahead of policy or pilot implementation and following public engagement j) Exploration of ethical issues – this process unearthed several ethical issues that should be explored in relation to both policy testing and implementation. Ethical considerations can impact on the acceptability of pilots as well as the potential costs of mitigating ethical issues, thus do not stand alone in assessing overall feasibility.

160 Given the number of concrete areas for possible further exploration discussed above, we recommend that all partners involved continue to reflect on the learning and are active in dissemination of the study findings.

10.5.3 Discussion The obstacles identified through the feasibility study pose significant challenges in progressing the testing of CBI in Scotland, and indeed the rest of the United Kingdom. The practical difficulties in implementing a pilot alongside our current benefits system would require substantial resources to address current IT and legislative constraints. The levers to address these barriers sit within the UK government and considerable political effort would be required in order to make the legislative and regulatory changes required. Given our feasibility findings, it is perhaps not surprising that other contemporary pilots, especially those in Europe and North America, do not test all the fundamental aspects of basic income. Conditionality, for example, is a well-established characteristic of most social security systems and often enshrined in legislation, thus making experimenting with unconditionality challenging. Given this challenge, it is perhaps unsurprising that despite significant international interest, there has been limited progress in implementing basic income policy.

This is not to suggest that there would not be benefit in piloting CBI in Scotland. Indeed, the current evidence base suggests that a policy such as CBI could be transformative in Scottish society. The nature of employment is changing and the growth of the ‘gig economy’, zero-hour contracts, and part time working78 impacts on precariousness across society. CBI has the potential to change the nature of life choices that people currently face in relation to work, caring, learning, and creating. Evidence from economic modelling summarised in Section 9 suggests that a CBI in Scotland could lead to a substantial reduction in poverty, child poverty and inequality, whilst there is the sparse but promising evidence from existing literature on improved health and wellbeing, education and some social outcomes. However, Section 9 also shows that CBI policies like those described in Section 5 would have substantial cost implications as well as the potential to lead to slower economic growth over the medium to longer term and reduced real incomes for the richest groups in Scotland. Any slower growth and redistribution of incomes may reduce the carbon footprint of the economy, although the nature and mechanisms of this are not well evidenced.

The potential trade-off between slower economic growth and income restraint for the richest groups in Scotland with reductions in poverty, inequality and potential improved outcomes across health, wellbeing, education and social outcomes is a discussion that is already in progress. Scotland has been a frontrunner in the discussion on Wellbeing Economies, with Scotland hosting the first meeting of the Wellbeing Economy Governments (WEGo) group in Glasgow in October 2017 followed by a policy lab in Edinburgh in May 2019,xc and subsequent TED talk by First Minister Nicola Sturgeon on the topic.xci In this speech the First Minister argued that a focus on policies that promote wellbeing, and not just economic growth, is important for governments in a world of growing division and inequality. This talk signalled an ambition to take a new approach in Scotland to focus on broader measurements of health and wellbeing, not just economic growth. The possibilities of a CBI as a policy response to poverty, inequality, precarious employment and a focus on promoting wellbeing could contribute to that debate.

Interest in inclusive economies that promote public wellbeing goes beyond national government xc More information on WEGo in Scotland can be found at https://www.gov.scot/groups/wellbeing- economy-governments-wego/ xci First Minister Nicola Sturgeon TED talk can be viewed at https://www.ted.com/talks/nicola_sturgeon_ why_governments_should_prioritize_well_being/up-next 161 and is being operationalised strategically and at a local level through local government and public health in Scotland. As set out in the Scottish Local Authorities’ Economic Development (SLAED) Group’s current Strategic Plan,xcii local authorities are in the position to apply in practical terms the concept of inclusive growth, as set out in Scotland’s Economic Strategy. Local authorities are demonstrating a commitment to integrating economic wellbeing more generally, thereby helping ensure that all of Scotland’s communities can benefit from economic success. In doing so, local authorities are developing strong linkages between economic development and wider priorities in Scotland at national and local levels such as addressing inequalities, child poverty, connectivity, climate change and improving outcomes, including through approaches such as Community Wealth Building and the role of ‘anchor institutions’. Local authorities are contributing to the delivery of a range of national performance framework outcomes, including ‘globally competitive, entrepreneurial, inclusive and sustainable economy, thriving and innovative businesses, with quality jobs and fair work for everyone’.xciii The public health community in Scotland also recognise that not everyone lives in the conditions which allow them to be as healthy as they can be.

It should be the case that the social, economic and physical environments in which they live help create health and wellbeing, but this is not the case. Whilst the NHS plays a vital role, wellbeing cannot be created and sustained by the NHS alone. To that end, a public health priority of “A Scotland where we have a sustainable and inclusive economy with equality of outcomes for all” was agreed in 2018.xciv This priority recognises that income inequality undermines opportunities for our more disadvantaged groups and individuals. Most health differences find their root cause in differences in wealth and income. Health is intrinsically linked to our ability to participate fully in society and having the resources and social connections to do so.

This feasibility study has outlined evidence to suggest that CBI could play a role in achieving these outcomes, however there are gaps in our current knowledge as to how people would respond to a basic income in Scotland. A pilot would provide an opportunity to test some of these responses. The feasibility study outlines a well-considered pilot model design which is reflective of the current social security context, alongside a rigorous evaluation plan. However, there are considerable challenges for a Scottish CBI pilot relating to funding and institutional interactions, public knowledge and acceptance and possible ethical issues, all of which would require significant political will and strategic effort in order to implement a robust pilot of CBI in Scotland. It is hoped that the findings of this study can help further inform debate on what kind of society and economy is desired in Scotland in the future, as well as helping identify actions to help achieve future ambitions in relation to basic income or social security reform.

xcii Available at http://www.slaed.org.uk/documents/SLAED-strategic-plan-2019-22.pdf xciii SLAED Strategic Plan 2019-2022, p7. xciv For more information on Public Health Priority 5 see https://publichealthreform.scot/media/1580/a- scotland-where-we-have-a-sustainable-inclusive-economy-with-equality-of-outcomes-for-all.pdf

162 References

1 Basic Income Earth Network (2020) ‘About basic income’, Basic Income Earth Network Website. Available at: https://basicincome.org/basic-income/ [Last accessed: 13th February 2020} 2 Basic Income Earth Network (2020) ‘About BIEN’, Basic Income Earth Network Website. Available at: https://basicincome.org/about-bien/#history [Last accessed: 10th January 2020] 3 RSA (2018) ‘Realising Basic Income Experiments in the UK’, Royal Society for encouragement of Arts, Manufactures and Commerce. Available at: https://www.thersa.org/discover/publications- and-articles/reports/realising-basic-income [Last accessed: 11th December 2019] 4 TUC (2019) Insecure Work. Available at: https://www.tuc.org.uk/sites/default/files/Insecure%20 Work%20Report.pdf [Last Accessed: 10th December 2019] 5 Taulbut M., Mackay D.F., McCartney G., Job Seeker’s Allowance (JSA) benefit sanctions and labour market outcomes in Britain, 2001–2014, Cambridge Journal of Economics, Volume 42, Issue 5, September 2018, Pages 1417–1434, https://doi.org/10.1093/cje/bex088 6 Standing G. (2017) Basic Income: And how we can make it happen’, Penguin Random House UK. 7 Lupton J. (2019) ‘As an artist, a Basic Income would transform my life’, Citizens’ Basic Income Network Scotland, 22 April. Available at: https://cbin.scot/2019/04/22/artist-basic-income- transform-my-life/ [Last accessed: 11th December 2019] 8 Brook O., O’Brien D. and Taylor M. (2018) Panic! Social class, taste and inequalities in the creative industries. Available at: https://createlondon.org/wp-content/uploads/2018/04/ Panic-Social-Class-Taste-and-Inequalities-in-the-Creative-Industries1.pdf [Last accessed: 10th December 2019] 9 Hiilamo, H. (2018) “Experimenting with basic income in Finland, Canada and the United States: What can we learn?” Session E1, BIEN Congress 2018. Tampere, Finland. 10 Gibson, M., Hearty, W. and Craig, P. (2018) Universal basic income: a scoping review of evidence on impacts and study characteristics. Project Report. What Works Scotland 11 Gibson, M., Hearty, W. and Craig, P. (2020) The public health effects of interventions similar to basic income: a scoping review. Lancet Public Health 2020; 5: e165-76. 12 Evans W.N., Moore T.J. The short-term mortality consequences of income receipt. Journal of public economics 2011; 95(11-12): 1410-24. 13 Widerquist, K. and A. Sheahen (2012). The United States: The Basic Income Guarantee — Past Experience, Current Proposals. Basic Income Worldwide: Horizons of Reform. M. C. Murray and C. Pateman. London, Palgrave Macmillan UK: 11-32. 14 Barclay, C., McLachlan, J., and Paterson, M. (2019). Exploring the practicalities of a basic income pilot. International Learning Report. Carnegie Trust UK. https://www.carnegieuktrust.org.uk/ publications/exploring-the-practicalities-of-a-basic-income-pilot/ 15 Wright, S. , Stewart, A. B.R. and Dwyer, P. (2018) Final findings: social security in Scotland. Project Report. Welfare Conditionality. 16 Kangas O., Jauhianinen S., Simanainen M. and Ylikännö M. (eds) (2019) The Basic Income Experiment 2017-2018 in Finland. Preliminary Results, available at: http://julkaisut.valtioneuvosto.

163 fi/bitstream/handle/10024/161361/Report_The%20Basic%20Income%20Experiment%20 20172018%20in%20Finland.pdf?sequence=1&isAllowed=y [last accessed 8th November 2019] 17 Santens S. (2019) What is there to learn from Finland’s Basic Income experiment? Did it succeed or fail? Medium, 14th February 2019, available at: https://medium.com/basic-income/what-is- there-to-learn-from-finlands-basic-income-experiment-did-it-succeed-or-fail-54b8e5051f60 [last accessed 6th March 2020] 18 Kangas O., Niemelä M. and Ylikännö M. (2019) All you need is trust! Results from the Finnish basic income experiment, University of Bath IPR Blog, 14th November 2019, available at: http:// blogs.bath.ac.uk/iprblog/2019/11/14/all-you-need-is-trust-results-from-the-finnish-basic-income- experiment/ [last accessed 29th November 2019] 19 Government of Ontario (2019) Archived – Ontario Basic Income Pilot, available at: https://www. ontario.ca/page/ontario-basic-income-pilot, page updated 8th April 2019 [last accessed 2nd December 2019] 20 Basic Income Canada Network (2019) Signposts to Success: Report of a BICN Survey of Ontario Basic Income Recipients, available at: https://assets.nationbuilder.com/bicn/pages/42/ attachments/original/1551664357/BICN_-_Signposts_to_Success.pdf [last accessed 2nd December 2019) 21 European Social Survey 2016 data and documentation available at: https://www. europeansocialsurvey.org/data/download.html?r=8 22 Ipsos MORI 2017 Universal Basic Income Research for University of Bath, further information available at: https://www.ipsos.com/sites/default/files/ct/news/documents/2017-09/omnibus- universal-basic-income-topline-2017.pdf 23 Populus on behalf of RSA Universal Basic Income Survey 2018, further information available at: https://www.thersa.org/globalassets/pdfs/reports/rsa-populus-basic-income-polling.pdf 24 Fife Council People’s Panel – Survey 28 June-July 2017, results available at: http://publications. fifedirect.org.uk/c64_Survey28StatsandCommentsBW1.pdf 25 North Ayrshire Council People’s Panel 2018 Summary Report available at: http://northayrshire. community/wp-content/uploads/2018/10/NorthAyrshirePeoplesPanelReport2018.pdf 26 Congreve, E., Connolly, K., Eiser, D., Gunson R., Kumar, A., McGregor, P., Roy, G., and Statham, R., 2020, “Modelling the Economic Impact of a Citizen’s Basic Income in Scotland”. https://www. sbs.strath.ac.uk/download/Fraser/202004/Modelling-Economic-Impact.pdf 27 Bregman, R. (2018) “Utopia for Realists.” Keynote Lecture, Nordic BI Day. Tampere, Finland. 28 Miller A. (2017) A Basic Income Handbook, Luath Press, Edinburgh 29 Barclay, C. (2017) Basic Income, Fife Council Research, Policy and Coordination Committee, September 2017 https://wordpress.fifedirect.org.uk/basicincomescotland/wp-content/uploads/ sites/75/2018/04/Appendix-Basic-Income.pdf 30 Painter A. and Thoung C. (2015) Creative Citizen, Creative State: The principled and pragmatic case for a Universal Basic Income, The RSA, https://www.thersa.org/globalassets/reports/rsa_ basic_income_20151216.pdf 31 Ross-Tatam J. (2017) A Secure Foundation to Build Our Lives: Making the case for Universal Basic Income (UBI), The Buchan Institute, https://docs.wixstatic.com/ugd/fda07c_ ed45429052b348d6aa46cb9ef449c0f4.pdf

164 32 Story M. (2015) Free Market Welfare: The case for a Negative Income Tax, Adam Smith Institute, https://static1.squarespace.com/static/56eddde762cd9413e151ac92/t/56f711a3ecb92886bb6 cc478/1459032484139/NIT_WEB.pdf 33 Reform Scotland (2016) The Basic Income Guarantee, https://reformscotland.com/wp-content/ uploads/2016/02/The-Basic-Income-Guarantee.pdf 34 Reed H. and Lansley S. (2016) Universal Basic Income: An idea whose time has come? Compass, https://www.compassonline.org.uk/wp-content/uploads/2016/05/ UniversalBasicIncomeByCompass-Spreads.pdf 35 Lansley S. and Reed H. (2019) Basic Income for All: From desirability to feasibility, Compass, http://www.compassonline.org.uk/wp-content/uploads/2019/03/Compass_ BasicIncomeForAll_2019.pdf 36 Torry M. (2017) ‘A variety of indicators evaluated for two implementation methods for a Citizens Basic Income’, Euromod Working Paper Series, https://www.euromod.ac.uk/sites/default/files/ working-papers/em12-17.pdf 37 Painter A., Thorold J. and Cooke J. (2018) Pathways to Universal Basic Income: The case for a Universal Basic Opportunity Fund, The RSA, https://www.thersa.org/globalassets/pdfs/reports/ rsa_pathways-to-universal-basic-income-report.pdf 38 Gunson R., Baxter D. and Stirling A. (2018) How much would it cost to reduce child poverty in Scotland? IPPR Scotland, https://www.ippr.org/files/2018-02/reducing-child-poverty-in-scotland- february18.pdf 39 Painter A., Cooke J., Burbidge I. and Ahmed A. (2019) A Basic Income for Scotland, The RSA, https://www.thersa.org/globalassets/pdfs/rsa-a-basic-income-for-scotland.pdf 40 Standing G. (2019) Basic Income as Common Dividends: Piloting a Transformative Policy, https://www.progressiveeconomyforum.com/wp-content/uploads/2019/05/PEF_Piloting_Basic_ Income_Guy_Standing.pdf 41 Wright, E. O., 2006. “Compass Points: Towards a Socialist Alternative.” New Left Review 41: 93- 124. 42 Wright, E. O., 2010. Envisioning Real Utopias. London: Verso 43 De Wispelaere, J. and Noguera J. A., “On the Political Feasibility of Universal Basic Income: An Analytic Framework”, in Caputo, R. (Ed) (2012), Basic Income Guarantee and Politics. Palgrave Macmillan US. 44 Goodman C. and Danson M. (Eds.) (2019) Exploring Basic Income in Scotland: Basic Income and Housing, available at: https://cbin.scot/wp-content/uploads/2019/04/Housing-Exploring- Basic-Income-in-Scotland.pdf 45 Forget, E. (2011) ‘The Town with No Poverty: The health effects of a Canadian Guaranteed Annual Income field experiment,’Canadian Public Policy 37(3) pp.283-305 46 Baumberg, B., Bell, K., Gaffney, D., with Deacon, R., Hood, C. and Sage, D. (2012)Benefits stigma in Britain, London: Elizabeth Finn Care/Turn2us, Available at: https://wwwturn2us-2938. cdn.hybridcloudspan.com/T2UWebsite/media/Documents/Benefits-Stigma-in-Britain.pdf 47 Taulbut M, Mackay DF, McCartney G. Job Seeker’s Allowance (JSA) benefit sanctions and labour market outcomes in Britain, 2001–2014. Cambridge Journal of Economics 2018; 42(5): 1417–1434, https://doi.org/10.1093/cje/bex088.

165 48 Pawson, R. and Tilley, N., Realistic Evaluation, Sage, 1997 49 Craig P et al. Developing and evaluating complex interventions. Medical Research Council, 2006. https://mrc.ukri.org/documents/pdf/complex-interventions-guidance/ 50 Craig P et al. Developing and evaluating complex interventions. Medical Research Council, 2006. https://mrc.ukri.org/documents/pdf/complex-interventions-guidance/ 51 McHugh, N., Biosca, O., Baker, R., Ibrahim, F. and Donaldson, C. (2019). Innovating on Methods to Understand the Relationship Between Finances and Wellbeing. In Wealth(s) and Subjective Well-Being, edited by Brulé G., Suter C. Social Indicators Research Series, vol 76. Springer, Cham. 52 Morduch, J. and Schneider, R. (2017). The financial diaries: How American families cope in a world of uncertainty, Princeton University Press: Princeton, New Jersey 53 Jowell, R. (2003) Trying it out: The role of ‘pilots’ in policy-making: Report of a review of Government pilots, Cabinet Office/Office of the Social Researcher’s Office: London.https:// assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/ file/498256/Trying_it_out_the_role_of_pilots_in_policy.pdf (Accessed 25 February 2020) 54 Craig, P., Katikireddi, S., Leyland, A. et al (2017) Natural Experiments: An overview of methods, approaches and contribution to public health intervention research, Annual Review of Public Health, 38, pp. 39-56 https://doi.org/10.1146/annurev-publhealth-031816-044327. 55 White, H. and Sabarwal, S. (2014) Quasi-experimental design and methods, Methodological Briefs: Impact Evaluation No. 8, Unicef Office of Research, Florencehttps://www.unicef-irc.org/ KM/IE/impact_8.php (Accessed 25 February 2020) 56 Craig, P., Dieppe, P., Macintyre, S. et al (2006) Developing and evaluating complex interventions: new guidance, Medical Research Council. www.mrc.ac.uk/ complexinterventionsguidance (Accessed 25 February 2020) 57 Verweij, M. (2009) Equipoise in public health research, in A. Dawson (ed) The philosophy of public health, Ashgate, Farnham, Surrey, pp. 145 – 154. 58 Petticrew, M., McKee, M., Lock, K. et al (2013) In search of social equipoise, BMJ 2013;346:f4016 doi: 10.1136/bmj.f4016 (4 July) 59 Social Security Advisory Committee (2007) An examination of the ethical and legal issues of “piloting” in its widest sense in the Department for Work and Pensions, SSAC https://assets. publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/323425/ ssac_occasional_paper_2.pdf (Accessed 25 February 2020) 60 Mackay, D. and Chakrabati, A. (2019) Government Policy Experiments and Informed Consent, Public Health Ethics, 12, 2, pp. 188-201 61 Kela (2016) From Idea to Experiment: Report on Universal Basic Income Experiment in Finland, Working Papers 106, Available at: https://helda.helsinki.fi/bitstream/handle/10138/167728/ WorkingPapers106.pdf (Accessed 25 February 2020) 62 BSA (2017) Statement of Ethical Practice, British Sociological Association https://www.britsoc. co.uk/media/24310/bsa_statement_of_ethical_practice.pdf(Accessed 23 July 2019) 63 Respect Code of Practice http://www.respectproject.org/code/respect_code.pdf. (Accessed 23 July 2019) 64 WMA (2013) Declaration of Helsinki: Ethical principles for medical research involving human

166 subjects https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for- medical-research-involving-human-subjects/ (Accessed 23 July 2019) 65 Sim, J. and Dawson, A. (2012) American Journal of Public Health,102, pp.480–485. doi:10.2105/ AJPH.2011.300389 66 Weijer, C., Grimshaw, J., Eccles, M. et al (2012) The Ottowa Statement on the ethical design and conduct of Cluster Randomized Trials, PLOS Medicine, 9, 11, pp.1-9, Nov. https://journals.plos. org/plosmedicine/article?id=10.1371/journal.pmed.1001346 (Accessed 26th February 2020) 67 Scottish Government (2012) Guiding Principles for Data Linkage, Scottish Government, Edinburgh https://www2.gov.scot/Topics/Statistics/datalinkageframework/GuidingPrinciples (Accessed 23 July 2019) 68 Inclusion Scotland (2020) Friend or foe? Basic income and disabled people in Scotland, available at: https://inclusionscotland.org/wp-content/uploads/2020/02/Basic-Income- discussion-paper.pdf 69 Shaw J. and Paterson J. (2019) Exploring the Social Security Implications of a Citizens Basic Income Pilot, Child Poverty Action Group in Scotland, https://basicincome.scot/wp-content/ uploads/sites/75/2019/06/Exploring_the_social_security-implications_of_a_basic_income_ pilot_March2019.pdf 70 SOLAR (2018) ‘Review of Local Governance SOLAR Appendix 2’, Publications: SOLAR Responses, Available at: http://www.solarscotland.org.uk/wp-content/uploads/2019/03/Review- of-Local-Governance-SOLAR-Appendix-2-V2.docx, Last Accessed 27th January 2020. 71 Painter A. and Thoung C. (2015) Creative Citizen, Creative State: The principled and pragmatic case for a Universal Basic Income, The RSA, https://www.thersa.org/globalassets/reports/rsa_ basic_income_20151216.pdf 72 Hirsch D. (2015) ‘Could a ‘Citizens Income’ Work?’ Joseph Rowntree Foundation Programme Paper, p. 5, https://www.jrf.org.uk/sites/default/files/jrf/migrated/files/citizens-income-full.pdf 73 Gunson R., Baxter D. and Stirling A. (2018) How much would it cost to reduce child poverty in Scotland? IPPR Scotland, https://www.ippr.org/files/2018-02/reducing-child-poverty-in-scotland- february18.pdf 74 Murray, C. (2016). In our hands: A plan to replace the welfare state. Washington DC, The AEI Institute. 75 Smith, M. and G. Shanahan. (2018). “Is a basic income the solution to persistent inequalities faced by women?” Retrieved 29 August 2019, from https://theconversation.com/is-a-basic- income-the-solution-to-persistent-inequalities-faced-by-women-92939 76 Gibson, M., H. Thomson, K. Banas, V. Lutje, M. J. McKee, S. P. Martin, C. Fenton, C. Bambra and L. Bond (2018). “Welfare-to-work interventions and their effects on the mental and physical health of lone parents and their children.” The Cochrane Library. 77 Layard, R., Nickell, S. Jackman, R., 2005. Unemployment: Macroeconomic Performance and the Labour Market. (Oxford University Press, Oxford.) 78 Fraser of Allander Institute & Scottish Centre for Employment Research. Scottish Labour Market Trends Vol 3, No 1.

167 www.basicincome.scot [email protected]

168